Pipeline & Hazardous Materials Safety Administration · 2018-03-23 · The PHMSA Pipeline Safety...

Pipeline & Hazardous Materials Safety Administration

Pipeline Inspection Technologies Demonstration Report

Pipeline Safety Research & Development

Program

Final

2

EXECUTIVE SUMMARY The pipeline infrastructure is a critical element in the energy delivery system across the United States. Its failure can affect both public health and safety directly and indirectly through impacts on the energy supply. The pipeline infrastructure is aging, while at the same time Research & Development (R&D) funding from the pipeline industry to develop technologies to assure its integrity is experiencing budgetary constraints. Total R&D funding is being further reduced through the elimination of programs resulting from restructuring within the government and energy industry. The Pipeline & Hazardous Materials Safety Administration (PHMSA), Pipeline Safety R&D Program mission is to ensure the safe, reliable & environmentally sound operation of the nation’s pipeline transportation system. With passage of the Pipeline Safety Improvement Act (PSIA) in 2002, industry is now required to invest significantly more capital to inspect and maintain their systems. The PSIA requires enhanced maintenance programs and continuing integrity inspection of all pipelines located within “high consequence areas” where a pipeline failure could threaten public safety, property and the environment. According to the Interstate Natural Gas Association of America (INGAA) the cost to industry to implement the PSIA in the first ten years will exceed $2 billion. The focus of the PHMSA Pipeline Safety R&D Program is to sponsor research and development projects intended on providing near-term solutions that will improve the safety, reduce environmental impact, and enhance the reliability of the nation’s pipeline transportation system. Conducting in-field technology demonstration test to facilitate technology transfer from government funded R&D programs strengthens communication and coordination with industry stakeholders The PHMSA Pipeline Safety R&D Program role in technology development and innovation has increased with the passage of the Pipeline Safety Improvement Act of 20021. The implementation of the Integrity Management Program for natural gas and hazardous liquids has focused efforts on proactively finding and fixing safety-related problems. For several years the PHMSA Pipeline Safety R&D Program along with the DOE/NETL, Gas Delivery Reliability Program have funded the development of advanced in-line inspection (ILI) technologies to detect mechanical damage, corrosion and other threats to pipeline integrity. Several projects have matured to a stage where demonstrations of their detection capability are now warranted. During the week of January 9th, 2006, the PHMSA Pipeline Safety R&D Program and the DOE/NETL, Gas Delivery Reliability Program co-sponsored a demonstration of six innovative technologies.

1http://www.eia.doe.gov/oil_gas/natural_gas/analysis_publications/ngmajorleg/pubsafety.html

The keys to enhanced pipeline safety are understanding the risks, focusing on the problems, imagining solutions, and applying our ingenuity—

Ted Willke.

3

The demonstrations were conducted at Battelle West Jefferson’s Pipeline Simulation Facility (PSF) near Columbus, Ohio. The pipes used in the demonstration were prepared by Battelle at the PSF and each was pre-calibrated to establish baseline defect measurements. Each technology performed a series of pipeline inspection runs to determine their capability to detect and size mechanical damage, corrosion, stress corrosion cracking or plastic pipe defects. Overall, each technology performed well in their assessment category. BACKGROUND Information regarding inspection technology advances needs to be disseminated and understood by many stakeholders in the pipeline industry. While research reports, review meetings and conference presentations are commonly used to disseminate information, live demonstrations can provide additional information on the current state and future potential of each development. Demonstrations are challenging to technology developers because newly developed technologies must be sufficiently reliable to obtain results in a fixed time frame. There is not the opportunity to return to the laboratory to confirm results or change parameters. While the pressure to demonstrate the best capability of their technology advances is enormous, the developers understand these events are needed to bolster support for continued development. The results of demonstrations can be difficult to directly compare since each implementation can be at a different stage of development. No direct comparisons were made in this report. At this demonstration, representatives from the pipeline industry, industry trade associations, and pipeline service providers were able to witness the performance of six new technologies and interact with technology developers to clarify the current and potential capability of these new developments. The participation of these groups was an essential element of the demonstration. This is the second benchmark of emerging pipeline inspection technologies performed by Battelle for DOT PHMSA Pipeline Safety R&D Program and DOE NETL. Information on the pipe defect sets, pipe preparation, demonstration facility layout, and demonstration procedures from the first test can be can be found in the final report, Benchmarking Emerging Pipeline Inspection Technologies2, prepared by Battelle. The results from the first benchmarking can be found in the Pipeline Inspection Technologies - Demonstration Report3, prepared by NETL. Purpose This report provides a brief summary assessment of the demonstration benchmark results. The purpose of this assessment is to help identify promising inspection technologies best suited for further development as part of an integrated teaming effort between robotic platform and sensor developers. This report is not intended to provide a detailed analysis of each technology’s performance or to rate their performance relative to one another.

2 http://primis.rspa.dot.gov/matrix/FilGet.rdm?fil=718 3 http://www.netl.doe.gov/technologies/oil-gas/publications/td/Battelle%20Inspection%20Demo%20Final%20Report_111804.pdf

4

The Technologies Six innovative sensor technologies were demonstrated at Battelle’s Pipeline Simulation Facility (PSF) the week of January 9, 2006. The different technologies demonstrated their ability to detect pipeline corrosion, mechanical defects, stress corrosion cracking, or plastic pipe defects. Additional information on each technology may be found in both Appendix B and Appendix C. The technologies were: ORNL Shear Horizontal Electromagnetic Acoustic Transducer (EMAT) – Oak Ridge National Laboratory (ORNL) has developed an EMAT system that uses shear horizontal waves to detect flaws on natural gas pipelines. A wavelet-based analysis of ultrasonic sensor signals is used for detecting physical flaws (e.g., SCC, circumferential and axial flaws, and corrosion) in the walls of gas pipelines. Using an in-line non-contact EMAT transmitter-receiver pair, flaws can be detected on the walls of the pipe that the current magnetic flux leakage (MFL) technology has problems detecting. One EMAT is used as a transmitter, exciting an ultrasonic impulse into the pipe wall while the second EMAT located a few inches away from the first is used as a receiving transducer. ORNL’s technology is depicted in Figure 1.

Figure 1. ORNL Shear Horizontal EMAT GTI Remote Field Eddy Current (RFEC) – The Gas Technology Institute (GTI) has developed a RFEC inspection technique to inspect pipelines with multiple diameters, valve and bore restrictions, and tight or miter bends. This electromagnetic technique uses a simple exciter coil that can be less than on third of the pipe diameter and is driven by a low-frequency sinusoidal current to generate an oscillating electromagnetic field that small sensor coils can detect. The oscillating field propagates along two paths; a direct axial path and an indirect or remote path.

5

The direct field attenuates rapidly because the pipe acts as a waveguide that will only allow frequencies in the gigahertz range and above to propagate. It becomes negligible after 2 to 3 pipe diameters. Thus after 2 to 3 pipe diameters, the only signal left is that from the remote field, which propagates out through the pipe wall, along its exterior and then re-enters the pipe 2 to 3 pipe diameters from the exciter coil. This is exactly what is needed for defect detection since the electromagnetic waves must now pass directly through metal loss defect regions and other flaws. Changes from nominal values of the amplitude and phase of the remote field detect defects in the pipe wall and measure their severity. GTI’s technology is depicted in Figure 2.

Figure 2. GTI Remote Field Eddy Current SwRI Remote Field Eddy Current (RFEC) – Through funding support from PHMSA/OPS, Southwest Research Institute® has developed a remote-field eddy current (RFEC) technology to be used in unpiggable lines. The SwRI RFEC tool is capable of detecting corrosion on the inside or outside pipe surface. Since a large percentage of pipelines cannot be inspected using “smart pig” techniques because of diameter restrictions, pipe bends, and valves, a concept for a collapsible excitation coil was developed but found unnecessary for the pipe sizes and materials of interest in this demonstration. A breadboard system that meets the size, power, and communication requirements for integration into the Carnegie Mellon Explorer II robot was developed and used in the demonstration tests. This system is shown in Figure 3. The demonstration system incorporates eight detectors, and data from all eight channels are acquired and processed simultaneously as the system is scanned along the pipe at speeds up to 4 inch/sec. All of the instrumentation, except for a DC power supply and a laptop computer (used for storage of the processed data), is located on the tool. The RFEC system can expand to inspect 6- or 8-inch-diameter pipe and can retract to 4 inches to pass through obstructions.

Sensor Coils

MUX Board

Mock Explorer Module

Drive Coil

Support

6

Laptop Computer with CAN Bus Interface

Encoder Wheel

Electronics Sensors Excitation Coil

DC Power Supply

Figure 3. SwRI Remote Field Eddy Current PNNL Ultrasonic Strain Measurement – Pacific Northwest National Laboratory (PNNL) has developed an ultrasonic sensor system capable of detecting pipeline stress and strain caused by mechanical damage i.e., dents and gouges. PNNL has established the relationship between residual strain and the change in ultrasonic response (shear wave birefringence) under a uniaxial load. Initial measurements on samples in both axial and biaxial states have shown excellent correlation between shear birefringence measurements. The demonstration focused on refining the methodology, particularly under circumstances when the damage is more complex than a simple uniaxial deformation. PNNL’s technology is depicted in Figure 4.

7

EMAT Sensor

Motor for sensor rotation

Springs for smooth motion past dents

Figure 4. PNNL Ultrasonic Strain Measurement Rotating Permanent Magnet – Battelle is developing a rotating permanent magnet inspection system where pairs of permanent magnets are rotated around the central axis. This alternative to the more common concentric coil method can be used to induce high current densities in the pipe. Along the pipe away from the magnets in either direction, the currents flow in the circumferential direction. Anomalies and wall thickness variations are detected with an array of sensors that measure local changes in the magnetic field produced by the current flowing in the pipe. The inspection methodology can be configured to pass tight restrictions and narrow openings such as plug valves. The separation between the magnets and the pipe wall is on the order of an inch (2.5cm). The strength of circumferential current produces signals on the order of a few gauss, which can be detected by hall effect sensors positioned between 8 and 40 inches (10 and 100 cm) away from the rotating magnets. This evolving inspection methodology was first demonstrated in summer of 2004. Battelle’s technology is depicted in Figure 5.

8

Figure 5. Battelle Rotating Permanent Magnet Capacitive Sensor for Polyethylene Pipe Inspection – The National Energy Technology Laboratory (NETL) has developed a capacitive probe to resolve defects in plastic natural gas pipelines. This new technology uses a non-destructive and non-hazardous projected electric field to map voids and other anomalies. The probe can function autonomously and is intended for use in conjunction with existing “pigs” or on its own platform. NETL’s technology is depicted in Figure 6.

Figure 6. NETL Capacitive Sensor

9

Demonstration Configuration The emerging inspection technologies were tested within a 40 by 100 foot high-bay area at Battelle’s PSF. Pipes selected for these tests had various types of natural and machined defects. A black tarp and bubble wrap covered the pipes to hide defect locations. Figure 7 shows the configuration of the pipes during the demonstration. These pipes included:

Figure 7. High-bay Looking North Detection of Metal Loss

• One 8-inch diameter ERW seam welded pipe measuring 30-feet in length (0.188 inch wall thickness). The pipe sample contained two rows of simulated corrosion defects spaced 180° apart.

• One 8-inch diameter ERW seam welded pipe measuring 35-feet in length (0.188 inch wall thickness). The pipe sample contained two rows of simulated corrosion defects spaced 180° apart. This sample also included a 5-foot section of natural corrosion from a pipe pulled from service.

• One 8-inch diameter ERW seam welded pipe measuring 40-feet in length (0.188 inch wall thickness). The pipe sample contained two rows of simulated corrosion defects spaced 180° apart.

10

Detection of Mechanical Damage

• One 24-inch diameter pipe measuring approximately 28-feet in length (0.292 inch wall thickness) comprised of two separate pipes welded together with mechanical damage defects. Three rows of mechanical damage defects were located on this pipe sample spaced 120° apart but only one row with track hoe defects were used in the benchmarking.

• One 24-inch diameter pipe measuring approximately 40 feet in length (0.292 inch wall thickness) with plain (or smooth) dent defects along one row.

Detection of Stress Corrosion Cracking (SCC)

• One 26-inch diameter pipe measuring approximately 26 feet in length (0.281 inch wall thickness) with natural stress corrosion cracking. A separate 26-inch diameter SCC pipe sample was provided for calibration.

Detection of Plastic Pipe Defects

• One 6-inch diameter polyethylene pipe measuring 13 feet in length (0.5 inch wall thickness) with cylindrical drill holes and saw cut defects along one row on the exterior of the pipe.

Additional information on the pipe defect sets, pipe preparation, demonstration facility layout, and demonstration procedures can be found in the final benchmarking report, Pipe and Anomaly Configuration for the Phase II Benchmarking of Emerging Pipeline Inspection Technologies prepared by Battelle and included in Appendix D. DEMONSTRATION RESULTS This section provides an assessment of the test data relative to the benchmark data developed at the Battelle Pipeline Simulation Facility (PSF). The benchmark data is provided as Appendix A of this document and test results for the individual technologies, as prepared and submitted by the technology developers, can be found in Appendix B. Metal Loss Corrosion Assessment The three corrosion assessment technologies were demonstrated in an 8-inch diameter pipe4. This diameter was chosen to match a specific crawler implementation, Explorer, being separately developed under NETL DOE and Northeast Gas Association (NGA) funding5. The untethered platform is designed to traverse pipelines ranging from 6 to 8 inches inside diameter. The inspection technology developers were asked to include as many of the configuration and interface requirements of this platform as practical. Three 8-inch diameter pipes were inspected by each technology for corrosion. The first pipe (Pipe Sample 1) was a seam-welded pipe measuring approximately 35 feet in length. This sample consisted of three pipe sections welded together (two circumferential welds) and

4 In the first demonstration these technologies were demonstrated in 12-inch diameter pipe. 5 http://www.netl.doe.gov/technologies/oil-gas/publications/td/41155_Final.PDF

11

contained simulated corrosion defects set along two test lines 180° apart. The simulated corrosion was created using electrochemical etching techniques, an example of which is shown in Figure 8. A 5 foot section of Pipe Sample 1 also contained natural corrosion from a pipe recently pulled from service.

Figure 8. Example Simulated Corrosion Defect using Electrochemical Etching Techniques The donated natural corrosion pipe sample had a field girth weld with corrosion on both sides of the weld. The weld drop through was too large for the inspection tool specifications and as such the pipe was trimmed to include roughly 2 feet of corrosion on one end, 3 feet of full thickness pipe at the other end, and no field welds. The pipe was then sandblasted and welded between two new pipes to comprise Pipe Sample 1. When the pipe was being fully characterized, an additional weld was found in the middle of the corrosion area (see Figure 9). This weld was very fine and did not have a significant crown. The natural corrosion defects were intended to be a “stretch goal” of these emerging inspection technologies. While the natural corrosion sample represents a real world problem, this additional weld adds a complex scenario that is most likely new to the technology developers. As such, these search areas are reported but are not included in the results evaluation.

12

Figure 9. Fine, Field Weld in Natural Corrosion Pipe Segment The second pipe (Pipe Sample 2) was a seam-welded pipe measuring approximately 30 feet in length. This sample consisted of two pipe sections welded together (one circumferential weld) and contained simulated corrosion defects set along two test lines 180° apart. The third pipe (Pipe Sample 3) was a seam-welded pipe measuring approximately 40 feet in length. This sample consisted of two pipe sections welded together (one circumferential weld) and contained simulated corrosion defects along two test lines 180° apart. All three technologies detected one false positive signal; however, none of the technologies had a false positive in the same location. None of the technologies failed to identify a defect and were fairly accurate in predicting the locations. These results are summarized in Table 1. In addition, the corrosion sizing results were plotted in a manner commonly used by pipeline inspection vendors to demonstrate commercial in-line inspection technology capabilities. For these graphs, benchmark data is plotted against the values reported by the technology developers. Care must be taken in interpreting these graphs since:

• Error in the benchmark measurements is not zero

• Only the maximum depth is compared while the corrosion pit depth varied throughout the defect; many corrosion areas had more than one area of local thinning.

• Length and width were measured at the surface; however other measures can also be used that still accurately describe the anomaly.

Overall these graphs show the results predicted by each technology correlated well with the benchmark data.

Field Weld

13

Table 1. Detection Rates for the Corrosion Technologies Technology Detection

Rate False Positive Rate False

Negative Rate

Mean Difference in Location of Defect

Standard Deviation of Defect Location

SwRI – RFEC 100% (32 of 32)

3.3% (1 of 30) – Defect P2-8 called as a repeatable signal, but does not have typical flaw signal characteristics; 0.17”

deep, 1.38” long and 1.06” wide

0% (0 of 32)

0.33 1.71

GTI – RFEC 100% (32 of 32)

3.3% (1 of 30) – Defect P1-2 called as an unknown feature resembling metal loss; 0.008” deep, <1” long, and >4” wide

0% (0 of 32)

0.08 1.18

Battelle – Rotating Permanent Magnet

100% (32 of 32)

3.3% (1 of 30) – Defect P1-17 called as a small single pit 0.02” deep, 0.7” long, and 0.75” wide

0% (0 of 32)

-0.31 2.05

SwRI Results SwRI began testing the morning of Monday, January 9, 2006, and completed testing by mid-day Thursday, January 12, 2006. The SwRI RFEC tool acquired, processed, and displayed data in real time as it was continuously pulled through each pipe sample. Each scan took approximately 5 minutes to complete with selected higher speed runs taking approximately one to two minutes to complete. A circumferential region of 60 degrees was inspected in each scan, and two scans were made along each defect line to ensure complete coverage of all defects. The SwRI RFEC technology detection rate was 100%, detecting all defect sites on Pipe Sample 1, Pipe Sample 2, and Pipe Sample 3. On average, SwRI located anomalies slightly past the actual start of the defect location with a standard deviation of 1.71 inches. The SwRI RFEC technology detected one false positive signal on Test Line 1 of Pipe Sample 2. The false positive signal was identified as a repeatable signal without typical flaw signal characteristics with a depth of nearly 90% of the wall thickness and approximately 1 5/8 -inch in length. SwRI’s sizing accuracy is depicted in Figures 10 through 12 in which the predicted and measured anomaly depths, lengths, and widths are presented.

14

Figure 10. Measured Depth vs. Predicted Depth for the SwRI RFEC

Figure 11. Measured Length vs. Predicted Length for the SwRI RFEC

0.0000

0.0188

0.0376

0.0564

0.0752

0.0940

0.1128

0.1316

0.1504

0.1692

0.1880

0.00

00

0.01

88

0.03

76

0.05

64

0.07

52

0.09

40

0.11

28

0.13

16

0.15

04

0.16

92

0.18

80

Measured Depth (inches)

Pred

icte

d D

epth

(inc

hes)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Measured Length (inches)

Pred

icte

d Le

ngth

(inc

hes)

15

Figure 12. Measured Width vs. Predicted Width for the SwRI RFEC GTI Results GTI began testing on the morning of Monday January 9, 2006 and completed testing by the evening of Thursday January 12, 2006. The GTI RFEC sensor technology collected data by indexing through each defect region in 0.25 inch steps. The GTI RFEC technology was able to scan both test lines in each pipe sample at the same time but because of the small incremental data collection each pipe sample required a full day to collect data. GTI did attempt a continuous scan with the results of this scan provided in Appendix C. The GTI RFEC technology detection rate was 100%, detecting all defect sites on Pipe Sample 1, Pipe Sample 2, and Pipe Sample 3. On average, GTI located anomalies slightly past the actual start of the defect location with a standard deviation of 1.18 inches. The GTI RFEC technology detected one false positive signal on Test Line 1 of Pipe Sample 1 but identified the anomaly as a small unknown feature with a depth of only 4% of the wall thickness and approximately 1-inch in length. GTI’s sizing accuracy is depicted in Figures 13 through 15 in which the predicted and measured anomaly depths, lengths, and widths are presented.

0

0.5

1

1.5

2

2.5

3

0 0.5 1 1.5 2 2.5 3

Measured Width (inches)

Pred

icte

d W

idth

(inc

hes)

16

Figure 13. Measured Depth vs. Predicted Depth for the GTI RFEC

Figure 14. Measured Length vs. Predicted Length for the GTI RFEC

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5


Pred

icte

d Le

ngth

(inc

hes)

0.0000

0.0188

0.0376

0.0564

0.0752

0.0940

0.1128

0.1316

0.1504

0.1692

0.1880

0.00

00

0.01

88

0.03

76

0.05

64

0.07

52

0.09

40

0.11

28

0.13

16

0.15

04

0.16

92

0.18

80


Pred

icte

d D

epth

(inc

hes)

17

0

0.5

1

1.5

2

2.5

3

3.5

4

0 0.5 1 1.5 2 2.5 3 3.5 4


Pred

icte

d W

idth

(inc

hes)

Figure 15. Measured Width vs. Predicted Width for the GTI RFEC Battelle Results Battelle began testing the afternoon of Tuesday January 10, 2006 and completed testing by the afternoon of Friday January 13, 2006. Battelle’s testing was periodically interrupted due to concerns from the other corrosion inspection technology developers that the permanent magnet was causing interference with their systems. The Battelle Rotating Permanent Magnet technology was able to continuously acquire data through each pipe sample taking approximately 10 to 15 minutes to scan one test line. During the demonstration Battelle processed signals and displayed inspection results in real-time. The Battelle Rotating Permanent Magnet technology detection rate was 100%, detecting all defect sites on Pipe Sample 1, Pipe Sample 2, and Pipe Sample 3. On average, Battelle located anomalies shy of the actual start of the defect location with a standard deviation of 2.05 inches. The Battelle Rotating Permanent Magnet technology detected one false positive signal on Test Line 2 of Pipe Sample 1 but identified the anomaly as a small single pit with a depth of only 11% of the wall thickness and approximately 3/4-inch in length. Battelle’s sizing accuracy is depicted in Figures 16 through 18 in which the predicted and measured anomaly depths, lengths, and widths are presented.

18

Figure 16. Measured Depth vs. Predicted Depth for the Battelle Rotating Permanent Magnet

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5


Pred

icte

d Le

ngth

(inc

hes)

Figure 17. Measured Length vs. Predicted Length for the Battelle Rotating Permanent Magnet

0.0000

0.0188

0.0376

0.0564

0.0752

0.0940

0.1128

0.1316

0.1504

0.1692

0.1880

0.00

00

0.01

88

0.03

76

0.05

64

0.07

52

0.09

40

0.11

28

0.13

16

0.15

04

0.16

92

0.18

80


Pred

icte

d D

epth

(inc

hes)

19

Figure 18. Measured Width vs. Predicted Width for the Battelle Rotating Permanent Magnet The benchmark data and test results for the three technologies that tested for metal loss on Pipe Samples 1, 2, and 3 are shown in Table 2 through Table 7.

0

0.5

1

1.5

2

2.5

3

0 0.5 1 1.5 2 2.5 3


Pred

icte

d W

idth

(inc

hes)

20

This page intentionally blank.

21

Table 2. Benchmark Data vs. Results for Corrosion Pipe Sample 1; Test Line 1

Simulated Corrosion Pipe Sample 1 Test Line 1 Defect Number P1-1 P1-2 P1-3 P1-4 P1-5 P1-6 P1-7 P1-8 P1-9 P1-10 P1-11 P1-12 Search Region (from End B) 328" to 340" 304" to 316" 280" to 292" 256" to 268" 232" to 244" 208" to 220" 184" to 196" 160" to 172" 120" to 144" 100" to 112" 76" to 88" 52" to 64"

Start and End of Defect (inches)5

Benchmark Data Blank Blank 287.75 290.875

259.625 263.625

232.75 235.75 Blank 190.625

192.75 Blank 120" 140.25" Blank Blank 56.75

60.875

SwRI –RFEC 282.6 285.8

254.5 258.7

227.7 231.0 188.8

189.7 160.0 172.0

a=120.0 122.3

b=128.5 129.3

56.9 60.2

GTI – RFEC ~311.25 ~314.25

~281.75 ~285.5

~260.5 ~264.75

~232 ~236.5 ~191.25

~193.75 ~120 ~134.25 ~56.75

~60.5 Battelle – Rotating Permanent Magnet 288

292.8 260.1 264.9

233.2 237 188.8

192.2 120 132 58.1

62.3 Defect Length (inches)

Benchmark Data Blank Blank 3.125 4 3 Blank 2.125 Blank 20.25 Blank Blank 4.125

SwRI –RFEC 3.16 4.20 3.30 0.95 12 a=2.25 b=0.77 3.32

GTI – RFEC < 1 2.875 3.375 3.75 1.625 14.25 2.875 Battelle – Rotating Permanent Magnet 3.8 3.8 2.8 2.4 12 3.2

Defect Width (inches) Benchmark Data Blank Blank 2 2 1 Blank 2 Blank Full Circ. Blank Blank 2

SwRI –RFEC 1.25 1.95 1.09 1.92 Full Circ. a=1.82 b=Full Circ. 1.63

GTI – RFEC > 4 1.5 ~3 > 3 1.5 >4 1.84 Battelle – Rotating Permanent Magnet 1.0 1.5 1.0 1.5 32 1.75

Maximum Defect Depth (inches) Benchmark Data Blank Blank 0.096 0.063 0.081 Blank 0.147 Blank 0.146 Blank Blank 0.122

SwRI –RFEC 0.10 0.06 0.08 0.09 0.18 a=0.066 b=0.83 0.13

GTI – RFEC 0.008 0.090 0.064

0.100 0.075 0.07 0.135

0.133 ~0.141 0.154 0.142

Battelle – Rotating Permanent Magnet 0.075 0.055 0.050 0.165 Various up to

0.150 0.115

Comments

SwRI –RFEC defect signal outside stated region

appears to be large region of general wall thinning that extends out of the designated region. Signal patterns not characteristic of calibration defects.

two defects in region, designated a and b.

GTI – RFEC unknown feature resembling metal loss, 4%

2 axially aligned pits, 48% and 34%

2 axially aligned pits, 53% and 40%

2 pits, deepest 37%. Additional features observed attributed to through hole of defect 18 sitting over drive coil

2 pits offset diagonally, 72% and 71% deepest pit was a single

small slit ~75% 2 axially aligned pits, 82% and 75.5%


corrosion patch, multiple pits of different depths




a slow change in signal in all sensor throughout the region indicates a material property change

large area of general corrosion of variable depth that spans the entire sensor width. The corrosion is close to the weld, altering both signals. A large wide corrosion area at 128"


22

Table 3. Benchmark Data vs. Results for Corrosion Pipe Sample 1; Test Line 2 Simulated Corrosion Pipe Sample 1 Test Line 2

Defect Number P1-13 P1-14 P1-15 P1-16 P1-17 P1-18 P1-19 P1-20 P1-21 P1-22 P1-23 Search Region (from End B) 330" to 342" 306" to 318" 282" to 294" 258" to 270" 234" to 246" 210" to 222" 186" to 198" 160" to 172" 120" to 144" 98" to 110" 74" to 86"

Start and End of Defect (inches)

Benchmark Data 335.75 339.625

308.875 312 Blank Blank Blank 213.625

217.875 Blank Blank 120 140.75

108 110

79.75 83.75

SwRI –RFEC 335.8 339.9

309.5 312.8 214.0

218.1 160.0 172.0

128.9 129.7

108.1 110.1

79.9 81.4

GTI – RFEC ~336.375 ~340.25

~309 ~312.75 214.75

218.875 ~120 ~134.25

~108.5 ~111

~79.75 ~83.5


334.2 338.4

306.9 310.8 241.1

241.8 212.5 216.8 126.0

138.0 103.1 106.3

74.8 79.1

Defect Length (inches) Benchmark Data 3.875 3.125 Blank Blank Blank 4.25 Blank Blank 20.75 2 4 SwRI –RFEC 4.04 3.31 4.13 12 0.79 1.99 1.48 GTI – RFEC 3 2.875 3.25 14.25 1.625 2.875 Battelle – Rotating Permanent Magnet 3.2 2.9 0.7 3.3 12 2.2 3.3

Defect Width (inches) Benchmark Data 1.75 1 Blank Blank Blank 2 Blank Blank Full Circ. 2 2 SwRI –RFEC 1.47 1.37 1.69 Full Circ. Full Circ. 1.82 1.72 GTI – RFEC 1.75 1.5 2 > 4 1.5 2.25 Battelle – Rotating Permanent Magnet 1.0 1.25 0.75 2.0 > 5 1.75 1.0

Maximum Defect Depth (inches) Benchmark Data 0.095 0.115 Blank Blank Blank 0.145 Blank Blank 0.127 0.12 0.097 SwRI –RFEC 0.08 0.11 0.14 0.18 0.06 0.08 0.09

GTI – RFEC 0.122 0.070

0.113 0.132 0.188

0.111 0.113 0.122 0.088 0.096

Battelle – Rotating Permanent Magnet 0.075 0.115 0.020 0.155 Various up to

0.150 0.110 0.075

Comments

SwRI –RFEC

appears to be large region of general wall thinning that extends out of the designated region. Signal patterns not characteristic of calibration defects.

GTI – RFEC 2 axially aligned pits, 65% and 37%

2 axially aligned pits, 70% and 60% 2 pits, through hole

and 59% general corrosion, deepest 60% and 65% diagonal feature, 47% 51%




small single pit corrosion patch, multiple pits of different depths

a slow change in signal in all sensor throughout the region indicates a material property change

area of general corrosion of variable depth that spans most sensors. A large wide corrosion area at 128"



23


Simulated Corrosion Pipe Sample 2 Test Line 1 Defect Number P2-1 P2-2 P2-3 P2-4 P2-5 P2-6 P2-7 P2-8 P2-9 P2-10 P2-11 Search Region (from End B) 294" to 306" 270" to 282" 246" to 258" 222" to 234" 198" to 210" 174" to 186" 150" to 162" 126" to 138" 102" to 114" 78" to 90" 54" to 66"


Benchmark Data Blank Blank Blank 227.25 229.375 Blank 180.25

183.375 153.125 156.375 Blank 108.125

112.25 80.125

84.5 Blank

SwRI – RFEC 227.6 229.8 180.2

183.5 153.4 156.7

129.8 131.1

108.2 112.3

80.0 84.3

GTI – RFEC 227.25 229.5 179.75

183 152.75 155.75 107.75

111.75 80.25

84

Battelle – Rotating Permanent Magnet 228.9 232.0 181.1

184.9 153.8 157.6 108.6

112.0 80.0 83.6

Defect Length (inches) Benchmark Data Blank Blank Blank 2.125 Blank 3.125 3.25 Blank 4.125 4.375 Blank SwRI –RFEC 2.21 3.23 3.31 1.38 4.05 4.31 GTI – RFEC 2.25 3.25 3 4 3.75 Battelle – Rotating Permanent Magnet 2.1 2.8 2.8 2.4 2.6

Defect Width (inches) Benchmark Data Blank Blank Blank 2 Blank 1 1 Blank 2 2 Blank SwRI –RFEC 1.57 0.99 1.18 1.06 2.14 1.88 GTI – RFEC 2 1 2 1.5 2.5 Battelle – Rotating Permanent Magnet 1.0 1.0 1.0 1.5 1.5

Maximum Defect Depth (inches) Benchmark Data Blank Blank Blank 0.079 Blank 0.114 0.085 Blank 0.158 0.147 Blank SwRI –RFEC 0.07 0.11 0.04 0.17 0.16 0.13 GTI – RFEC 0.037 0.073 0.026 0.142 0.188 Battelle – Rotating Permanent Magnet 0.075 0.075 0.065 0.165 0.170

Comments

SwRI –RFEC

Repeatable signal, but does not have typical flaw signal characteristics.

GTI – RFEC


Corrosion patch, with multiple pits of different depths



Corrosion patch, with large multiple pits of different depths

Corrosion patch, with large multiple pits of different depths

24


Simulated Corrosion Pipe Sample 2 Test Line 2 Defect Number P2-12 P2-13 P2-14 P2-15 P2-16 P2-17 P2-18 P2-19 P2-20 Search Region (from End B) 246" to 258" 222" to 234" 198" to 210" 174" to 186" 150" to 162" 126" to 138" 102" to 114" 78" to 90" 54" to 66"


Benchmark Data 248.125 250.25 Blank 202.625

205.75 Blank Blank 130 134.125 Blank Blank 57.75

60.875

SwRI –RFEC 248.1 249.8 202.3

205.4 129.1 133.2 56.3

59.7

GTI – RFEC 249 251 201

204 129.5 133.5 57.5

60.5


209.0 132.8 137.4 59.4

63.2

Defect Length (inches) Benchmark Data 2.125 Blank 3.125 Blank Blank 4.125 Blank Blank 3.125 SwRI –RFEC 1.72 3.10 4.14 3.37 GTI – RFEC 2 3 4 3 Battelle – Rotating Permanent Magnet 2.0 3.1 3.6 2.8

Defect Width (inches) Benchmark Data 2 Blank 1 Blank Blank 2 Blank Blank 1 SwRI –RFEC 1.21 1.21 1.69 1.25 GTI – RFEC 1.5 1.5 2 1.5 Battelle – Rotating Permanent Magnet 1.5 0.75 1.5 1.0

Maximum Defect Depth (inches) Benchmark Data 0.14 Blank 0.105 Blank Blank 0.112 Blank Blank 0.188 SwRI –RFEC 0.11 0.08 0.11 0.16 GTI – RFEC 0.148 0.081 0.159 0.176 Battelle – Rotating Permanent Magnet 0.115 0.075 0.105 0.180

Comments SwRI –RFEC GTI – RFEC Battelle – Rotating Permanent Magnet




Corrosion patch, with multiple pits of different depths, One pit may be through hole

25


Simulated Corrosion Pipe Sample 3 Test Line 1 Defect Number P3-1 P3-2 P3-3 P3-4 P3-5 P3-6 P3-7 P3-8 P3-9 P3-10 P3-11

Search Region (from End B) 384" to 396" 360" to 372" 330" to 342" 300" to 312" 270" to 282" 222" to 234" 186" to 198" 162" to 174" 138" to 150" 102" to 114" 66" to 78"


Benchmark Data Blank Blank 335 337.25

305.625 306.375

275 277.25 Blank 189.875

194 Blank 143.665 144.335

106.375 109.625 Blank

SwRI –RFEC 336.3 338.5

306.8 307.6

276.0 278.3 190.5

194.8 144.1 144.8

106.7 109.9

GTI – RFEC 335 337

306.5 307.75

275.75 277.75 190

193.5 143.5 144.75

106 109

Battelle – Rotating Permanent Magnet 338.0 341.0

307.8 309.3

276.4 279.5 187.8

193.0 144.2 145.5

108.8 112.4

Defect Length (inches) Benchmark Data Blank Blank 2.25 0.75 2.25 Blank 4.125 Blank 0.67 3.25 Blank SwRI –RFEC 2.19 0.78 2.29 4.22 0.73 3.19 GTI – RFEC 2 1.25 2 3.5 1.25 3 Battelle – Rotating Permanent Magnet 2.0 1.5 2.1 4.2 1.3 2.6

Defect Width (inches) Benchmark Data Blank Blank 2 0.75 2 Blank 2 Blank 0.67 1 Blank SwRI –RFEC 1.8 0.88 1.93 1.64 0.63 1.29 GTI – RFEC 1.5 1 2 2 1 2 Battelle – Rotating Permanent Magnet 1.75 0.75 1.5 1.5 0.5 1.5

Maximum Defect Depth (inches) Benchmark Data Blank Blank 0.133 0.148 0.103 Blank 0.115 Blank 0.120 0.156 Blank SwRI –RFEC 0.09 0.11 0.09 0.10 0.09 0.15

GTI – RFEC 0.164 0.158 0.142 0.119 0.173 0.148

0.112 0.182 0.176

Battelle – Rotating Permanent Magnet 0.165 0.150 0.105 0.105 0.080 0.160 Comments

SwRI –RFEC

GTI – RFEC Three pits Three pits Two pits axially aligned



Single Pit



Single Pit


26


Simulated Corrosion Pipe Sample 3 Test Line 2 Defect Number P3-12 P3-13 P3-14 P3-15 P3-16 P3-17 P3-18 P3-19 P3-20 P3-21 P3-22 P3-23

Search Region (from End B) 390" to 402" 356" to 368" 330" to 342" 306" to 318" 282" to 294" 248" to 260" 210" to 222" 180" to 192" 156" to 168" 126" to 138" 102" to 114" 66" to 78"


Benchmark Data 392.25 396.375 Blank 335.875

336.625 Blank Blank 250.625 253.75

214.5 217.625

185.765 186.485 Blank 130

134.125 Blank 69.5 73.625

SwRI –RFEC 392.5 396.6 336.9

337.7 251.6 254.8

215.2 218.4

186.4 187.2 130.5

134.5 69.5 73.8

GTI – RFEC 393 396 335.75

336.75 251.25 254

214.5 217.25

185.75 186.75 130.25

133.75 70.25 74


338.0 250.5 254.4

214.1 218.1

184.9 186.2 128.3

132.7 65.9 70.6

Defect Length (inches) Benchmark Data 4.125 Blank 0.75 Blank Blank 3.125 3.125 0.72 Blank 4.125 Blank 4.125 SwRI –RFEC 4.18 0.72 3.21 3.18 0.81 3.97 4.3 GTI – RFEC 3 1 2.75 2.75 1 3.5 3.75 Battelle – Rotating Permanent Magnet 3.6 0.9 2.9 3.0 1.3 3.4 3.7

Defect Width (inches) Benchmark Data 2 Blank 0.75 Blank Blank 1 1 0.72 Blank 2 Blank 2 SwRI –RFEC 1.69 0.51 0.79 1.10 0.73 1.48 1.85 GTI – RFEC 2.5 1 1.5 1.5 1 2 2 Battelle – Rotating Permanent Magnet 1.25 0.5 1.0 1.0 0.75 1.25 1.5

Maximum Defect Depth (inches) Benchmark Data 0.094 Blank 0.154 Blank Blank 0.07 0.091 0.139 Blank 0.103 Blank 0.088 SwRI –RFEC 0.08 0.13 0.06 0.09 0.11 0.10 0.09

GTI – RFEC 0.135 0.102 0.149

0.145 0.089 0.066

0.128 0.094

0.142 0.124 0.121

0.114 0.124 0.009

Battelle – Rotating Permanent Magnet 0.065 0.105 0.085 0.080 0.105 0.085 0.055 Comments

SwRI –RFEC

GTI – RFEC Two pits axially aligned

There was an increase in amplitude in this region. We concluded that the increase in the field was caused by the drive coil being located at P3-14. An actual defect may be "buried" in the field but it is not obvious.

Two pits axially aligned

Two pits axially aligned Two pits Reflection

from defect 10 Two features



Single Pit


Corrosion patch, with

multiple pits of different depths

Single Pit



27

Mechanical Damage Assessment Only one technology, the PNNL Ultrasonic Strain Measurement technology, was tested for assessment of mechanical damage. Two 24-inch diameter pipes were inspected by PNNL for mechanical damage. The first pipe (Pipe Sample 1) consisted of two pipes welded together with mechanical damage defects along three rows separated by 120° and measured approximately 28-feet in length. The test line on Pipe Sample 1 consisted of mechanical damage created using a 50-ton track hoe. An example mechanical damage defect from Pipe Sample 1 is shown in Figure 19. The second pipe (Pipe Sample 2) measured approximately 40 feet in length with plain (or smooth) dent defects along one test line. An example mechanical damage defect from Pipe Sample 2 is shown in Figure 20.

Figure 19. Example Mechanical Damage Defect from Pipe Sample 1

Figure 20. Example Mechanical Damage Defect from Pipe Sample 2

28

The benchmark data and test results for PNNL’s PNNL Ultrasonic Strain Measurement technology are shown in Table 8 and Table 9. Table 8. Benchmark vs. Test Results for Mechanical Damage Pipe Sample 1

PIPE SAMPLE 1 Dent Severity

0 = No damage 1 = Least Severe

2 = Moderately Severe 3 = Severe

4 = Most Severe

Defect Number

Search Region

(from End A)

Benchmark PNNL

Comments

D1 64.5” 2 4 dent start at 63.5" end at 66.5", length 3" D2 68.5” 2 2 dent start at 67" end at 70", length 3" D3 77.5” 2 1 dent start at 77.3" end at 78.1", length 0.9"

D4 105” 2 2.5 long dent along the axis dent start at 99.9" end at 110.4", length 10.5"

D5 114” 1 2.5 long dent along the axis dent start at 113.8" end at 119.9", length 6.1"

D6 162” Not Part of Benchmark

detected, damage looks as significant as a 3 or 4, length approximately 6 inches long or two dents approximately 2 inches long separated by 1 inch

195.5”

damage detected; damage looks as significant as a 3, (1 defect approximately 7 inches long or 2 defects, one 4 inches long and a second 2 inches long separated by approximately 1 inch)

D7 230” 2 3 dent start at 228.1" end at 234.7", length 6.7" D8 240” 1 2 dent start at 236.4" end at 242.1 length 5.7" D9 246” 1 0.5 dent start at 245.7" end at 246.2", length 0.5"

D10 267.5” 4 4 similar to calibration defects dent start at 264" end at 270.1", length 6.1"

D11 274” 2 4 similar to calibration defects dent start at 271" end at 276.1 length 5.1"

D12 280.5” 3 4 similar to calibration defects dent start at 277.3" end at 282.6", length 5.3"

D13 305.5” 4 NR out of scan range D14 310” 4 NR out of scan range D15 313” 3 NR out of scan range

29

Table 9. Benchmark vs. Test Results for Mechanical Damage Pipe Sample 2

PIPE SAMPLE 2 Dent Severity

0 = No damage 1 = Least Severe

2 = Moderately Severe 3 = Most Severe

Defect Number

Search Region

(from End A)

Benchmark PNNL

Comments

R03 109.25" 1 1 small degree of damage, start of dent 107" end 110" length 3"

R04 144” 3 2 moderate damage, start of dent 140.25" end 147.5", length 7.25"

R05 183” 2 3 significant damage, start of dent 178.75" end 187.25, length 8.5

R06 217” 1 1 small degree of localized damage, start of dent 215.5" end 220, length 4.5"

R07 253” 2 2 moderate damage, start of dent 250.5" end 258.5, length 8"

R08 289.5” 3 3 significant damage, start of dent 286.75 end 295.25, length 8.5"

R09 325” 2 2 moderate damage, start of dent 323" end 331", length 8" R10 360.5” 3 3 significant damage, start of dent 359" end 367", length 8" R11 397” 0 0 no dent

The term “dent severity” is used in this report to describe relative severity of dents within a specific pipe sample. The absolute severity of each dent is not known. Determining the severity of mechanical damage is difficult since there are no standards such as those used for corrosion anomalies. The criteria used to establish the benchmark severity ratings could differ from PNNL’s severity criteria and as such may have led to the discrepancies. PNNL began testing the afternoon of Monday January 10, 2006 and completed testing by the afternoon of Friday January 13, 2006. The PNNL Ultrasonic Strain Measurement technology only assesses the relative severity of mechanical damage defects. Location of dents is more practically performed by caliper tools and as such was not part of the evaluation criteria for this technology. Additionally, because PNNL was only required to identify dent severity at a specific location the scan speed was also not assessed. PNNL’s technology performed well on the mechanical damage sample with plain dents (Pipe Sample 2). There was discrepancy between the PNNL data and the benchmark at defect sites R04 and R05 on Pipe Sample 2; however the remaining defect locations correlated well. There were a number of differences between the benchmark data and the PNNL data for Pipe Sample 1. PNNL noted that the multiple dents and the non-circular nature of the pipe from the three rows of dent defects on Pipe Sample 1 created a significant amount of background deformation and thus stress and strain within the pipe sample. Due to these factors, the PNNL Ultrasonic Strain Measurement technology was not optimized for the degree of background deformation and is possibly the reason for the discrepancies between the benchmark data and PNNL’s results. PNNL indicated that additional tests would be desirable to help classify the dent severity for Pipe Sample 1.

30

Stress Corrosion Cracking Only one technology, the ORNL Shear Horizontal EMAT, was tested for detection of stress corrosion cracking. ORNL began testing the afternoon of Tuesday January 10, 2006 and completed testing by mid-day Thursday January 12, 2006. The ORNL Shear Horizontal EMAT technology acquired data as their inspection tool was continuously pulled through the pipe sample at the rate of about an inch per second. ORNL took multiple scans through each line to assess the consistency of the signal. Results were not displayed in real time; rather ORNL post processes the captured data to develop final results. ORNL claims post processing is minimal and could easily be performed during data acquisition with current generation computing power. As shown in Table 10 the technology ran three lines on a 26-inch diameter pipe with natural stress corrosion cracking. The EMAT technology detected one false positive signal on each test line. The configuration of the SCC defects could have contributed to the false positive readings. Because the EMAT configuration scans a minimum of 9-inches of the pipe’s circumference, some of the false positives could be the result of other cracks located in close proximity to the SCC defects under evaluation. Only one defect site (SCC2) provided no discernable signal; however magnetic particle analysis showed that these cracks are small and difficult to detect. Additionally, the location of the crack colony listed as SCC3 is off by a couple of inches. This is possibly due to defect (18), not considered as part of the test and located approximately 3-inches away in the circumferential direction, which may have been detected over the smaller SCC colony in SCC3. The most significant cracks (SCC8, SCC9, and SCC10) in the test sample were detected by the ORNL Shear Horizontal EMAT technology. An example SCC defect is shown in Figure 21. The benchmark data and test results for ORNL’s Shear Horizontal EMAT technology are shown in Table 10.

Figure 21. Example SCC Defect

31

Table 10. Benchmark vs. ORNL Test Results; SCC Testing

Benchmark ORNL Defect

Number

Search Region

(from End B)

Start of Crack

Region (from End B)

End of Crack Region (from

End B) Type of SCC

Start of Crack Region (from

End B)

End of Crack Region (from

End B) Type of SCC

Test Line 1

SCC1 242" to 254" Blank

SCC2 (5 & 4)

226" to 242" 225.25 238.25 Isolated

SCC3 (8)

210" to 222" 209.25 212.25 Colony 214 216 Isolated


SCC5 140" to 152" Blank 145 148 Colony; another

isolated at 142 Test Line 2


SCC7 234" to 246" Blank 236 237 Isolated

SCC8 (6)

210" to 222" 210.75 213.5 Colony 210 211 Isolated

SCC9 (7)

188" to 200" 189.25 193.5 Colony 194 196 Colony

SCC10 (9)

140" to 152" 141.5 145.5 Colony 144 149

Colony; looks like gap in the middle;

may be 2 sets separated by 1-inch.

Test Line 3

SCC11 (16)

225" to 245" 224.25 241.25 Colony 237 239

Isolated; After scanning, we

documented large dirt patches along line 3 We believe

EMATs lifted off the surface due to dirt

inside pipe. Reliability of data in

this area is low



SCC14 140" to 152" Blank 139 141 Isolated

Polyethylene Pipe Defects Only one technology, the NETL Capacitive Sensor for Polyethylene Pipe Inspection, was tested for detection of plastic pipe defects. This technology inspects for small volumetric anomalies with an NETL specified detection threshold of approximately 0.015 cubic inches. The measurement technology is localized and therefore anomalies in close proximity and pipe end effects do not influence its detection capabilities. A measure of defect significance was established based on the calibration defect which was 3/8-inch in diameter and 50% deep (0.028 cubic inches). The volume of the calibration defect was set at a significance of one. The significance of all other defects was based on the volume of the

32

calibration defect. An example defect is shown in Figure 22. This defect was calculated to have a volume of 0.04 cubic inches which equals a significance of 1.43. As shown in Table 11, the technology ran one test line on a 6-inch diameter polyethylene pipe sample.

Figure 22. Example Plastic Pipe Defect

33

Table 11. Benchmark vs. NETL Test Results; Plastic Pipe Testing

Benchmark NETL Defect

Number Search Region

Defect Location from Side A (to

center)

Significance of Defect (volume

ratio from calibration defect)

Defect Volume

Defect Diameter

Defect Location from

Side A (to center)

Significance of Defect (volume

ratio from calibration defect)

Defect Volume Comments

inches inches ratio in3 inches inches ratio in3

D1 21" to 27" 25” 1.57 0.044 0.375” 25.06” 1.38 0.039

For significance: defect calibration hole @ 18” = 1 Vol @ 18” = 0.028, Vol @ 25.06 = 0.039

D2 28" to 34" Blank None D3 35" to 41" Blank None D4 42" to 48" 46” 0.79 0.022 0.25” 45.62” 0.99 0.028 Volume = 0.028

D5 49" to 55" 53” 0.89 0.025 1/8” wide 1” long saw cut 52.55” 1.31 0.037 Volume = 0.037

D6 56" to 62" Blank None D7 62" to 70" 67” 1.57 0.044 0.375” 66.36” 1.15 0.033 Volume = 0.033 D8 70" to 76" Blank None D9 77" to 83" Blank None

D10 84" to 90" 88” 0.61 0.017 0.25” 87.15” 0.43 0.012 Volume = 0.012 D11 91" to 97" Blank None

D12 98" to 104" 102” 1.43 0.04 1/8” wide 1” long saw cut 101.03” 1.61 0.045 Volume = 0.045

D13 105" to 111" 109” 1.43 0.04 0.75” 107.84” 0.71 0.02 Volume = 0.020 D14 112" to 118" 116” 0.54 0.015 0.375” 114.75” 0.57 0.16 Volume = 0.016

D15 119" to 125" 123” and 123.5” 0.61 (each) 0.017 (each) 0.25” (each) 121.89” 0.74 0.74 Volume = 0.021

D16 126" to 132" Blank None?

Indications that a consistent amount of material may have been removed along entire length

D17 132" to 138" Blank None?

Indications that a consistent amount of material may have been removed along entire length

D18 138" to 144" 140” 1.25 0.035 0.75” 138.3” 1.13 0.032 Volume = 0.032 D19 144" to 150" 148” 1.11 0.031 0.75” 146.76” 0.71 0.020 Volume = 0.020

Not part of the benchmarking demonstration

34


35

While this was the second demonstration for all other technology developers, this demonstration was the first for the NETL Capacitive Sensor technology and should be taken into consideration when evaluating the results. During the demonstration, the NETL Capacitive Sensor technology collected data at a frequency of 1-hertz but has the capability to collect data up to a frequency of 45-hertz. NETL’s accuracy in assessing defect severity is depicted in Figure 23. The NETL Capacitive Sensor technology detection was excellent detecting all defect sites to within 1% of the actual centerline location and did not report any false positive signals. The percentage difference in defect significance was approximately 25%.

0.0000

0.2000

0.4000

0.6000

0.8000

1.0000

1.2000

1.4000

1.6000

1.8000

0.00

00

0.20

00

0.40

00

0.60

00

0.80

00

1.00

00

1.20

00

1.40

00

1.60

00

1.80

00Measured Significance

Pred

icte

d Si

gnifi

canc

e

Figure 23. Measured Severity vs. Predicted Severity for the NETL Capacitive Sensor SUMMARY Four pipeline anomaly conditions were evaluated by six different sensor technology developers. Three technologies assessed corrosion anomalies while individual technologies assessed mechanical damage, SCC, and plastic pipe material loss. The corrosion detection techniques demonstrated significant promise for inspection of unpiggable pipelines. Accurate detection and sizing of natural corrosion appears to be reachable

36

but additional development may be required to refine sizing algorithms especially when pipe material properties are unknown and calibration defects are not available. Additional data processing for some of the technologies and collection of larger natural corrosion defect libraries to conduct repeatable testing needs to be established. Future collection of data towards target corrosion on pipe samples pulled from service will improve system capabilities. In addition, the speed at which data is collected could be improved for all of the technologies. The usability of these technologies will rely on their ability to collect data for long pipeline segments in a relatively short amount of time as well as their ability to meet the design and power requirements of the Explorer robotic platform. PNNL’s mechanical damage detection technique also achieved reasonably good results especially in the pipe sample containing only plain dents. Considering the uniqueness of Pipe Sample 1 (multiple dents in close proximity), more accurately assessing the dent severity for this type of pipe sample would be a future goal for PNNL’s technology. In-service pipelines with the amount of denting evident on Pipe Sample 1 is highly unlikely and does not represent a realistic pipeline operating scenario. Track hoe defects; however, would be typical of third party damage evident on operating pipelines. The ORNL EMAT system also performed well detecting natural stress corrosion cracks that formed while the pipeline was in-service. The ORNL EMAT technology did detect some false positives on each test line but was also able to detect the most significant SCC locations. Given the nature of SCC, it is difficult to accurately size crack depths. Some of the cracks used in the benchmarking program may have been too small to clearly detect. Collection of additional SCC defect libraries and crack sizing would be a valuable addition to this benchmarking program. The NETL Capacitive Sensor was quite accurate in identifying defect locations. Sizing of plastic pipe defects is reachable but will require additional research to develop defect sizing algorithms. While this was a successful demonstration of the inspection sensor technology, inspection variables need to be considered in future evaluations. Following the submittal of their test data, the technology developers were sent the benchmark data. They were given an opportunity to comment on their results and to provide their perspective on their technology’s performance relative to the benchmark data. Appendix C contains the developer’s comments. Overall, the technologies performed well and the results are encouraging. As the development of these technologies progresses and future testing takes place, it is envisioned that improvements in the technology and data analysis techniques will continue to improve the false positive rate and enhance the precision and accuracy of the defect signals. PATH FORWARD As noted, PHMSA Pipeline Safety R&D Program goals are to understand the gaps between existing technologies and those needed to resolve the key pipeline issues. One recognized path forward is to integrate successfully demonstrated sensor technologies into a robotic platform/sensor system that can be deployed remotely as part of an integrated package. This effort is driven in large part by new PSIA regulations which require inspection of gas

37

transmission pipelines and distribution mains in high-consequence areas. A large percentage of these pipes cannot be inspected using typical “smart-pig” techniques because of diameter restrictions, pipe bends and valves. In addition, pressure differentials and flow can be too low to push a pig through some pipes. To help solve these problems, the PHMSA Pipeline Safety R&D Program has established an aggressive schedule to develop a prototype remote system which includes continued co-funding with industry partners. It is anticipated that upon completion of the prototype systems, they will be able to traverse all pipes (including unpiggable lines) of various diameters while providing continuous, real-time detection of pipe anomalies or defects.

38


APPENDIX A – BENCHMARK DATA

A-1

Calibration Metal Loss Location

Depth of Metal LossMeasured Length &

Width of DefectMeasured Max. Depth of Defect

Comments

inches from End B to center of defect inches

361" (59" from End A) See profile

Defect Number

Search Region (Distance from End B)

Start of Metal Loss Region from Side B

End of Metal Loss Region from Side B

Total Length of Metal Loss Region Width of Metal Loss RegionMaximum Depth of Metal Loss Region

Additional Data Attached?

Comments

inches inches inches inches inches inches Y/N

Defect Number






Comments


BLANK 7

Defect 8

Defect 7

N

Y

Y

BLANK 10

Defect 9

BLANK 9

P1-13 330" to 342" 335.75"

BLANK 6

BLANK 5

BLANK 1

Defect 11

Defect 4

Defect 5

BLANK 3

Defect 10

P1-NC2

BLANK 11 (natural corrosion pipe segment)

Defect 3

Defect 2

BLANK 2

Y P1-NC1

BLANK 4 (natural corrosion pipe segment)

Y

Y

N

Y

N

---

WELD 120"

339.625" 3.875" 1.75" 0.095"

P1-14 306" to 318" 308.875" 312" 3.125" 1" 0.115"

Y

N

N

N

N

Y

Y

--- ---

WELD 180"

P1-1 328" to 340" --- ---

---

290.875" 3.125"

---

3"

2.125"

20.25"

--- ---

WELD 120"

P1-22

WELD 180"

---

--- --- ---

--- --- N BLANK 8

---

4.125"

---P1-11 76" to 88" --- ---

2"P1-12 52" to 64"

---

P1-10 100" to 112" --- ---

P1-15 282" to 294" --- ---

P1-16 258" to 270" --- ---

--- --- N

---

P1-17 234" to 246" --- ---

4.25" 2" 0.145" YP1-18 210" to 222" 213.625" 217.875"

--- --- --- NP1-19 186" to 198" --- ---

--- --- --- NP1-20 160" to 172" --- ---

20.75" Full Circumference 0.127" YP1-21 120" to 144" 120" 140.75"

98" to 110" 108" 110" 2" 2" 0.12" Y

0.122"

0.063"

2" 0.096"

1"

--- ---

56.75" 60.875" Defect 6

P1-2 304" to 316" --- --- --- --- ---

P1-5

P1-3 280" to 292" 287.75"

P1-4 256" to 268" 259.625"

2" 0.147"

263.625" 4" 2"

0.081"

Detection of Metal Loss - Page 1

Sensor Design:

Name:Date:Company:

TEST LINE 1

TEST DATA

8" Diameter, 0.188" Wall Thickness Pipe Sample; Schedule 10; Length = 34' 11.75"Pipe Sample:Defect Set:

PIPE SAMPLE 1

Calibration P1-1:

Metal Loss Length & Width

inches

2 x 2

CALIBRATION DATA

PIPE SAMPLE 1:

Pipe Sample

0.146"

---

P1-9 120" to 144" 120" 140.25"

--- ------ ---

Full Circumference

---

P1-8

P1-6 208" to 220" ---

P1-7 184" to 196" 190.625" 192.75"

160" to 172"

232" to 244" 232.75" 235.75"

P1-23

TEST LINE 2

74" to 86" 79.75" 83.75" 4" 2" 0.097"

A-2




Comments


301.5" (58.5" from End A) See profile275" (85" from End A) See profile

Defect Number






Comments


Defect Number






Comments


WELD 120"

Y

Defect 11; through hole

BLANK 11

BLANK 10

Defect 10

BLANK 9

BLANK 8

Defect 9

BLANK 7

Defect 8

BLANK 1

Y

N

N

N

TEST LINE 2

--- --- ---

60.875"

BLANK 3

BLANK 2

Defect 4

BLANK 5

Defect 3

Defect 2

Defect 1

Y

N

P2-9 102" to 114" 108.125"

P2-7

Detection of Metal Loss - Page 2Name:Date:Company:

Benchmarking of Inspection Technologies

250.25"

WELD 120"

BLANK 4

2.125" 2" 0.14"P2-12 246" to 258" 248.125"

P2-13 222" to 234" --- --- N--- --- ---

3.125" 1" 0.105" YP2-14 198" to 210" 202.625" 205.75"

--- --- --- NP2-15 174" to 186" --- ---

--- --- --- NP2-16 150" to 162" --- ---

4.125" 2" 0.112" YP2-17 126" to 138" 130" 134.125"

P2-18 102" to 114" --- ---

P2-19 78" to 90" --- --- --- --- ---

TEST LINE 1

Sensor Design:

--- --- ---

3.125" 1" 0.188"P2-20 54" to 66" 57.75"

0.158"112.25" 4.125" 2"

--- --- ---

150" to 162" 153.125" 156.375"

P2-8 126" to 138" --- ---

PIPE SAMPLE 2:Calibration P2-1: 3 x 1Calibration P2-2: 2 x 2

CALIBRATION DATA

Pipe Sample Metal Loss Length & Width

inches

3.25" 1" 0.085"

0.114"3.125"

Y

Y

P2-5 198" to 210" ---

TEST DATA

Pipe Sample: PIPE SAMPLE 2Defect Set:

---

P2-6 174" to 186" 180.25" 183.375"

--- ---

1"

--- N

P2-4 222" to 234" 227.25" 229.375" 2.125" 2" 0.079" Y

P2-3 246" to 258" --- --- --- --- --- N

P2-2 270" to 282" --- --- --- --- --- N

P2-11 54" to 66" --- ---

P2-1 294" to 306" --- ---

P2-10 78" to 90" 80.125" 84.5" Defect 54.375" 2" 0.147" Y

8" Diameter, 0.188" Wall Thickness Pipe Sample; Schedule 10; Length = 30' 0.375"

BLANK 6--- --- --- N

A-3




Comments


421" (59" from End A) See profile

Defect Number






Comments


Defect Number






Comments


Defect 8

WELD 240"

WELD 240"

N

Y

BLANK 10

Defect 13

BLANK 9

BLANK 6

Defect 12; machined defect

Defect 11

Defect 10

BLANK 8

BLANK 7

BLANK 1

Y

N

Y

N


BLANK 4

Defect 5

BLANK 3

Defect 14

Defect 4


Defect 2

BLANK 2

4.125" 2" 0.094"

0.75" 0.75"

P3-12 390" to 402" 392.25" 396.375"

0.154" N Defect 9; machined defectP3-14 330" to 342" 335.875" 336.625"

P3-15 306" to 318" --- --- --- --- --- N

--- --- --- NP3-16 282" to 294" --- ---

3.125" 1" 0.07" YP3-17 248" to 260" 250.625" 253.75"

3.125" 1" 0.091" YP3-18 210" to 222" 214.5" 217.625"

0.72" 0.72" 0.139" NP3-19 180" to 192" 185.765" 186.485"

--- --- --- NP3-20 156" to 168" --- ---

4.125" 2" 0.103" YP3-21 126" to 138" 130" 134.125"

--- --- ---

4.125" 2" 0.088"

--- ---

P3-22 102" to 114" --- ---

P3-23 66" to 78" 69.5" 73.625"

---384" to 396" --- ---

P3-11 66" to 78"

TEST LINE 2

--- --- --- N BLANK 5--- ---

--- --- --- NP3-2 360" to 372" --- ---

2.25" 2" 0.133" YP3-3 330" to 342" 335" 337.25"

0.75" 0.75" 0.148" NP3-4 300" to 312" 305.625" 306.375"

2.25" 2" 0.103" YP3-5 270" to 282" 275" 277.25"

--- --- --- NP3-6 222" to 234" --- ---

4.125" 2" 0.115" YP3-7 186" to 198" 189.875" 194"

--- --- --- NP3-8 162" to 174" --- ---

P3-9 138" to 150" 143.665" 144.335" 0.67" 0.67" 0.120" N

Defect Set: 8" Diameter, 0.188" Wall Thickness Pipe Sample; Schedule 10; Length = 40' 0.25"

TEST DATA

Pipe Sample: PIPE SAMPLE 3


Company:

Sensor Design:

P3-13 356" to 368" --- --- --- --- ---

Detection of Metal Loss - Page 3Name:Date:

1" 0.156"

inchesPIPE SAMPLE 3:

TEST LINE 1

Defect 7

Calibration P3-1: 2 x 2

P3-1

CALIBRATION DATA

Pipe Sample Metal Loss Length & Width

P3-10 102" to 114" 106.375" 109.625" 3.25"

A-4


A-5

A-6


A-7

A-8

Pipe Sample: 993Calibration

Crack Location

Length DepthMeasured

LengthMeasured

Depth

inches from end B inches

% wall thickness

186.4 2.558.7 586.4 582.4 2.544.4 3

Defect Number

Search Region (Distance from

End B)

Start of Crack Region from

Side B

End of Crack Region

from Side B

inches inches inchesIsolated CrackColony of Cracks

Isolated CrackColony of Cracks


Isolated CracksColony of Cracks


TEST LINE 1

SCC2(5 & 4)

226" to 242" 225.25 238.25None

SCC1(Blank 3)

242" to 254" --- ---None

NoneBlank 1

SCC3(8)

210" to 222" 209.25 212.25None

SCC5 (Blank 1)

140" to 152" --- ---

multiple cracks; max = ~1/2"multiple cracks; max = ~1/2"

23

Blank Area:

Comments

26" Diameter Pipe with Stress Corrosion Cracks; Length = 26 feetPipe Sample:

Comments

1 multiple cracks; max = ~3/4"

Defect Set:893

Benchmarking of Inspection TechnologiesDetection of SCC - Page 1

Sensor Design:

Name:

Date:Company:

CALIBRATION DATA

TEST DATA

multiple cracks; max = ~1/4"multiple cracks; max = ~3 1/4"

SCC4(Blank 2)

175" to 187" --- ---None

Type of SCC

45

Blank 2

Multiple 1/4" cracks; cracked area 2 3/4" by 2 1/2"

Two isolated cracks; cracked area 4" by 1 1/2" with ~2" long crack; cracked area 5 1/4" by 1 1/4" with ~3" long crack

Blank 3

A-9

Defect Number


End B)


Side B

End of Crack Region

from Side B






Sensor Design:

Type of SCC Comments

TEST DATA

Pipe Sample: 893

TEST LINE 2

None

141.5 145.5None

None

Blank


Name:

Defect Set: 26" Diameter Pipe with Stress Corrosion Cracks; Length = 26 feet

Date:Company:

SCC6(Blank 5)

246" to 258" --- ---

188" to 200" 189.25 193.5

SCC8(6)

210" to 222" 210.75 213.5None

SCC7(Blank 4)

234" to 246" ---

Blank

---None

Multiple cracks; max ~1/4" long; cracked area 3 1/2" by 3 1/2"

Multiple cracks; max ~1/4" long; cracked area 4 1/4" by 3 3/4"

Multiple cracks; max ~1/2" long; cracked area 3" by 2 1/2"

SCC10(9)

140" to 152"

SCC9(7)

A-10

Defect Number


End B)


Side B

End of Crack Region

from Side B





TEST LINE 3

None

140" to 152" --- ---

SCC13(Blank 7)

188" to 200" --- ---

None

Date:Company:

SCC14(Blank 6)

Blank

Blank

Blank

SCC12(Blank 8)

210" to 222" --- ---None

NoneMultiple cracks; max ~3/4" long; cracked area 17" by 1 3/4"SCC11

(16)225" to 245" 224.25 241.25


Name:

Sensor Design:

Type of SCC Comments

TEST DATA

Pipe Sample: 893Defect Set: 26" Diameter Pipe with Stress Corrosion Cracks; Length = 26 feet

A-11

Calibration Defect Location

Volume of Defect Depth of Defect Diameter of Defect

inches from end A cubic inches inches inchesC1: 18 0.028 0.25 0.375

Defect Number

Search Region (Distance from End

A)

Location of Defect Region from Side A

Significance of Defect (based on volume ratio

from calibration defect)

Volume of Defect (in3) (provided to

participant after defect signif reported)

Depth of Defect (in) (provided to participant

after defect signif reported)

Diameter of Defect (in) (provided to


Comments

inches inches

Calibration Defect = 1Less Severe <1More Severe >1 cubic inches inches inches

----

0.75

0.75

Comments

0.25

----

0.125

----

0.25

0.125

----

BLANK 0 ---- ----

53" 0.89

D6 56" to 62" BLANK

49" to 55"

D2

D4 42" to 48" 46"

D3 35" to 41" BLANK 0

28" to 34" BLANK 0 ----

D1 21" to 27" 25" 1.57 0.044 0.4

---- ----

0.375

CALIBRATION DATA

Defect

LINE 1

TEST DATA

6" Diameter, 0.5" Wall Thickness Pipe Sample, ~13' in lengthPipe Sample:Pipe Parameters:

PLASTIC PIPE SAMPLE

Benchmarking of Inspection TechnologiesDetection of Plastic Pipe Defects - Page 1

Sensor Design:

Name:Date:Company:

---- ----

0.450.022

0.2

0 ---- ----

0.025

0.79

D5

D7 63" to 69" 67" 1.57 0.044 0.4

Saw Cut ~1" long and 1/8" wide

Same as D1

D8 70" to 76" BLANK 0 ---- ----

0.375

----

----D9 77" to 83" BLANK 0 ----

D12 98" to 104" 102" 1.43 0.04 0.35 Saw Cut ~0.9" long and 1/8" wide

D11 91" to 97"

0.04 0.09D13 105" to 111" 109" 1.43 0.75

0.015 0.14D14 112" to 118" 116" 0.54 0.375

0.017 (each) 0.35 (each) Defect consists of two identical holes 1/2" apartD15 119" to 125" 123" and 123.5" 0.61 (each) 0.25 (each)

D16 126" to 132" BLANK 0 ----

D17 132" to 138" BLANK 0

D19 144" to 150" 148" 1.11

D10 84" to 90" 88" 0.61

0.031 0.07

0.017 0.35

---- ----

---- ----

D18 138" to 144" 140" 1.25 0.035 0.08

----

A-12


APPENDIX B – DEMONSTRATION TEST DATA

B-1

SOUTHWEST RESEARCH INSTITUTE (SWRI) DEMONSTRATION TEST DATA

B-2


B-3




Comments


359 (59 from End A) See profile

298.5 (58.5 from End A) See profile277 (85 from End A) See profile

(59 from End A) See profile

Defect Number






Comments


Defect Number






Comments


No indication

309.5 N

N

No indication

No indication

No indication

No indication

No indication

Appears to be large region of general wall thinning that extends out of the designated region. Signal patterns are not characteristic of the

calibration defects.

No indication

N

Defect type signal outside stated region.

Two defects in region, designated a and b.

Appears to be large region of general wall thinning that extends out of the designated region. Signal patterns are not characteristic of the

calibration defects.

No indication

N

N

N

WELD 120"

312.8

P1-13 330" to 342"335.8 339.9 4.04 1.47 0.08

3.31P1-14 306" to 318" 1.37 0.11

N

N

N

N

WELD 180"

P1-1 328" to 340"

0.95 1.92

a=2.25, b=0.77 a=1.82, b=Full Circ.

WELD 120"

P1-10 100" to 112"

WELD 180"

Calibration P3-1:

3.32

P1-11 76" to 88"

1.63

2 x 2

3 x 12 x 2

P1-15 282" to 294"

P1-16 258" to 270" No indication

P1-17 234" to 246"

4.13 1.69 0.14 NP1-18 210" to 222" 214.0 218.1

P1-19 186" to 198"

0.06 N

P1-20 160" to 172"

160.0 172.0 12.00 Full Circ. 0.18 N

P1-21 120" to 144" 128.9 129.7

P1-22 98" to 110" 108.1 110.1

0.13P1-12 52" to 64" 56.9 60.2

P1-2 304" to 316"

0.06

P1-5

P1-3 280" to 292" 282.6 285.8 3.16 1.25 0.10

1.99 1.82 0.08 N

0.79 Full Circ.

0.09

258.7 4.20 1.95

1.09

189.7

Benchmarking of Inspection TechnologiesDetection of Metal Loss - Page 1

Sensor Design:

Name:Date:Company:

Gary Burkhardt27-Jan-06Southwest Research Institute

RFEC

TEST LINE 1

TEST DATA

Calibration P2-1:


PIPE SAMPLE 1

2 x 2

CALIBRATION DATA

PIPE SAMPLE 1:

PIPE SAMPLE 3:

Pipe Sample

PIPE SAMPLE 2:

Calibration P2-2:

Calibration P1-1:

Metal Loss Length & Width

inches

a=0.066, b=.083

12.00

P1-9 120" to 144" a=120, b=128.5 a=122.3, b=129.3

Full Circ. 0.18160.0 172.0

P1-8

P1-6 208" to 220"

P1-7 184" to 196" 188.8

160" to 172"

232" to 244" 227.7 231.0 0.083.30

P1-4 256" to 268" 254.5

P1-23

TEST LINE 2

74" to 86" 79.9 81.4 1.48 1.72 0.09

B-4

B-5

B-6


B-7

Comments on Tests Performed During Demonstration at Battelle: “Phase II Benchmarking Emerging Pipeline Inspection Technologies”

January 9–13, 2006

APPLICATION OF REMOTE-FIELD EDDY CURRENT (RFEC) TESTING TO INSPECTION OF UNPIGGABLE PIPELINES

OTHER TRANSACTION AGREEMENT DTRS56-02-T-0001 SwRI

® PROJECT 14.06162

PIPELINE AND HAZARDOUS MATERIALS SAFETY ADMNISTRATION U.S. DEPARTMENT OF TRANSPORTATION

SOUTHWEST RESEARCH INSTITUTE

®

January 2006

Demonstration tests of the remote-field eddy current (RFEC) method for inspection of 8-inch pipe were performed by Southwest Research Institute

® (SwRI

®). The target application of the

inspection technology is to integrate it with the Explorer II robot under development by Carnegie Mellon University. Therefore, the approach taken by SwRI was to perform the demonstration using a tool that meets the requirements and specifications for the Explorer II robot. All of the instrumentation (except for external power, which will be supplied by the robot), including excitation signal generation, amplification, filtering, multiplexing, analog-to-digital conversion, and digital signal processing (to provide phase-sensitive signal detection), was located on the RFEC tool. Total power required was less than half of the power budget available from the robot. Communication of commands and transfer of the processed signal data to an external computer were accomplished using a CAN bus—the same bus that will be used on the robot. Although the tool incorporated 8 channels (coverage of 60 degrees circumferentially) instead of the 48 channels intended for the robot tool (to achieve 360 degrees coverage), the circuitry is readily scalable to the full number of channels. Data were acquired by all 8 channels simultaneously during a single scan. The scans were made at a velocity of 1.5 inches/sec, and it was demonstrated that 4 inches/sec (the maximum scan speed of the robot) was possible. The data were post-processed for analysis to determine defect characteristics (length, width, and depth) using software that is readily adaptable to field inspections

The development of hardware that meets constraints associated with factors such as scan speed, power, and size always results in compromises that are not factors if, for example, laboratory instrumentation is used and if scan speeds are very slow. For example, slow scan speeds mean that significantly greater noise-reduction filtering can be used because time constants can be very long compared to those necessitated by fast scan speeds. Laboratory instrumentation can incorporate additional filtering and signal processing that cannot readily be performed by circuitry that must meet size and power constraints. Since the SwRI tool met the robot constraints, it can be expected that results similar to those achieved in these tests can be expected from the final integrated hardware.

B-8

It should be noted that defect characterization has a strong subjective element. In this demonstration, we were working with a brand new system, looking at defect types we had not seen before. That meant we had to use our best judgment and understanding of the RFEC method to interpret the indications. After the system has been used more extensively, experience will allow the operator to know quickly what type of defect is being detected based on the signal characteristics. The quantitative interpretation of the signals will then be improved over the present level. For example, the natural corrosion region in the demonstration pipes gave a signal unlike any of the calibration defects in our lab or supplied by Battelle. Furthermore, the signal extended beyond the designated region. As a result, we used our best judgment and reported the wall loss indicated by our depth algorithms. Magnetic field effects or the simple nature of RFEC response to very large area defects could cause our estimate to be in error. Familiarity with this type defect over a period of time would assure us of making a quicker and potentially more accurate appraisal of the corrosion.

B-9

GAS TECHNOLOGY INSTITUTE (GTI) Demonstration Test Da

Defect Number






Comments


Defect Number






Comments


WELD 120"

None

None

None

None

None

None

TEST LINE 2

None

None

None

None

None

Detection of Metal Loss - Page 2Name:Date:Company:


WELD 120"

Sensor Design:

P2-18 102" to 114"

P2-19 78" to 90"

P2-4

2 1.5 0.148

3 1.5 0.081

P2-12 246" to 258" 249 251

P2-13 222" to 234"

P2-14 198" to 210" 201 204

P2-15 174" to 186"

P2-16 150" to 162"

4 2 0.159P2-17 126" to 138" 129.5 133.5

3 1.5 0.176P2-20 54" to 66" 57.5 60.5

TEST LINE 1

P2-9 102" to 114" 107.75 0.142111.75 4 1.5

P2-7 150" to 162" 152.75 155.75

P2-8 126" to 138"

8" Diameter, 0.188" Wall Thickness Pipe Sample; Schedule 10; Length = 30' 0.375"

1

3 2 0.026

0.0733.25

P2-5 198" to 210"

TEST DATA

Pipe Sample: PIPE SAMPLE 2Defect Set:

P2-6 174" to 186" 179.75 183

222" to 234" 227.25 229.5 2.25 2 0.037

P2-3 246" to 258"

P2-2 270" to 282"

P2-11 54" to 66"

P2-1 294" to 306"

2.5 0.188P2-10 78" to 90" 80.25 84 3.75

B-10

B-11

A n a l y s i s o f S e n s o r B e n c h m a r k i n g T e s t s

Remote Field Eddy Current Technique

Prepared by: Julie Maupin, Albert Teitsma, Paul Shuttleworth

Gas Technology Institute 1700 S. Mount Prospect Road

Des Plaines, Illinois 60018

27 January 2006

B-12

Abstract During the week of 9 January 2006, GTI staff travelled to the Battelle Lab’s West Jefferson facility in Columbus, OH to test a prototype RFEC inspection vehicle in 3 samples of 8” pipe. We report briefly on the apparatus and its design, the electronic readout and data acquisition, and the analysis of the data. Where appropriate, we have discussed effects which lead to uncertainties in the location and size of reported defects. We also discuss uncertainties which may affect whether a defect would have been observable by our apparatus. Introduction The remote field eddy current (RFEC) technique is an electromagnetic, through-wall inspection technique for detecting defects and wall thinning in pipe walls. A simple exciter coil can be driven with a low frequency sinusoidal current to generate an oscillating magnetic field that small sensor coils can detect. This low frequency (10’s of Hz) oscillating field will propagate via two paths. It will propagate directly down the pipe a short distance. It will also propagate out through the wall, along the exterior of the pipe, and will re-enter the pipe --- the so-called indirect field. At axial distances of 2-3 pipe diameters from the exciter coil, the indirect field re-entering the interior of the pipe is much larger than the direct field coming from the exciter coil. Since it passes through the pipe wall, the indirect field contains information regarding the condition of the pipe. Changes from nominal value of the amplitude and phase of the indirect field indicate defects in the wall.

Figure 1: Paths of Energy Flow in the RFEC Technique. The remote field re-entering the pipe is the one containing the information regarding the condition of the pipe wall. We constructed a vehicle (“jig”) for carrying the RFEC apparatus. Near its front end it carried a solenoidal exciter coil, approximately 4” in diameter and 5” in length. It was comprised of1300 windings of 26 gauge wire. The sensor coils are located at distances of approximately 17” upstream of the exciter coil. They are ¾” in diameter, 3/8” in width, and contain approximately 20K windings of 50 gauge wire. Configured on the jig as two sets of 8 sensor coils, each set covered an angle of approximately 60º circumferentially at ¼” spacing. Mechanical Design The RFEC vehicle was composed of three parts, front support, rear support, and the center body. The front and rear supports had steering mechanisms on the wheels that helped keep the device upright and prevented any major rotation of the vehicle. The supports were coupled to

B-13

the center body, which contained all the equipment necessary to the RFEC technique. A picture of the center body is shown in Figure 2.

Figure 2: Center body of RFEC vehicle. GTI used two sets of 8 sensor coils to measure two defect lines simultaneously. The coils were mounted on shafts that served as pegs to attach the coils to plastic guides as shown in Figure 3. The guides were rounded to match the circumference of the pipe and routed on the leading edge to avoid jamming the welds. The guides were held against the pipe wall by spring-loaded, parallelogram configured arms. An end view of the sensor coil mounts is shown in Figure 4.

Sensor Coils

MUX Board

Mock Explorer Module

Drive Coil

Support

Plastic Coil Guide

Coil

Shaft

Figure 3: Diagram of sensor coils mounted to plastic guides.

Direction Of Travel

B-14

Figure 4: Sensor coil mounts inside an 8” pipe. The drive coil has been placed between two support modules, one having been built to imitate a module on the Explorer II robot. These support modules were important to keeping the drive coil centered in the pipe. GTI used an automatic winch system to pull the vehicle through the pipe. A tether line was attached to the front end of the vehicle. The tether wraps once around an encoder and then is wound onto a motor. The system is mounted directly onto the pipe and is controlled by LabVIEW to move the vehicle in ¼” steps. Uncertainties Related to Mechanics The jig suffered from some rotation inside the pipe. Each coil could have experienced rotations of up to ±10°. There were some encoder losses. After traveling 25’ in the pipe, we were measuring about 5” short of the actual location of the sensor coils. We eventually attached a fiberglass tape measure to the back end of the vehicle so we could always double check the encoder readings. In order to get good wall coverage from the coils, they had to be staggered, meaning half were closer to the drive coil than the other half. We have made provisions to correct the offset in the data analysis but there will still likely be an effect on the results.

Mounting

Spring

Plastic Coil

B-15

Electronics and Data Acquisition (DAQ) System GTI’s embodiment of a Remote Field Eddy Current inspection system is as follows: Signal Recovery 7265 DSP lock-in amplifier, Kepco BOP36-6M excitation coil driver, ADG407 16 channel multiplexer, Ni GPIB+ Gpib board and Ni PCI-6601 Counter Timer board. The preceding hardware is controlled by a Dell Pentium 4 workstation running at 2.99MHz with 1Gb of main memory and executing Lab View 7.1 under Windows XP Professional operating system. A general schematic of the DAQ system is shown in Figure 7. Channel addressing and distance gauging is accomplished using a Ni 6601 time/counter PCI circuit board. Distance measurements are made using a relative incremental encoder having a resolution of 1/16”.

Figure 5: Schematic of DAQ System. This figure schematically shows a 4-channel system. The system we operated at Battelle was a 16-channel version of this schematic. GTI’s RFEC machine is using a 100 count per revolution quadrature encoder. The encoder is interfaced to the system using a National Instrument PCI-6601 counter/timer circuit board. This circuit board supports 5 encoders; the encoder interface is done in hardware. The counter chip used in the NI circuit board has 32 bit registers giving a counting range of 268,435,453 inch. Data Collection Three LabView programs were used to collect data from the instrumentation on the jig. One read the encoder, one controlled the motor, and the other controlled the lock-in amplifier and acquired data from the coils. Acquiring the phase angle and magnitude of each coil was achieved by using a sequence of binary addressing to the multiplexer board. The program cycles through each coil sequentially. Once the data has been acquired for all 16 coil channels, the motor program fires the motor until the encoder program realizes it has traveled to the next ¼” step. Once the motor stops, the coils are again read and the phase and magnitude data is recorded to Excel. The process repeats. The lock-in amplifier has a programmable time constant for the low pass filter at its output. The program was written so that the operator could set the number of time constants that the program would wait at each coil address. Having a wait of multiple time constants ensured that

B-16

unsettled data would be flushed out and the readings would be accurate. The drawback to waiting for a certain number of time constants is slower acquisition time. It takes a significantly longer time to obtain data for 16 coils making overall inspection speed slow. No problems were encountered with LabVIEW. Analysis

Pipe Sample 3 Analysis of defect depth on Pipe Sample 3 was primarily done using Russell NDE Systems Inc.’s Adept Pro program. This program is the result of decades of research and focuses on the Voltage Plane for analysis. The display produced by the program is shown for Defect Line1 in Figure 6.

B-17

Figure 6: Adept Pro display of Defect Line 1 from Pipe Sample 3.

Weld

Defectsϕ

B-18

The display shows a strip chart of the phase angle on the left, followed by a C-Scan of the phase. Although the C-Scan provides a good overview of the defects, often as in this case, the strip chart is better for seeing the smaller defects. The magnitude information (strip chart and C-scan) is displayed to the right of the phase information. The top right hand panel shows the Voltage Plane. The black spiral is the attenuation spiral: as the wall thickness increases, the remote filed eddy current signal strength decreases while the phase also decreases, resulting in a spiral polar plot. The blue curve on the plot is the signal from the defect at the horizontal marker that runs across the strip charts and C-scans. The two red lines on either side of the marker delimit the range of data analyzed. If the blue line is extended to intersect the wall-thinning spiral, the vector from the origin of the polar plot to the intersection point makes an angle φ with the x-axis. Angle φ is used to determine the depth of the defect. The length of the blue line is used to find the circumferential extent of the defect. As in Figure 6, Figure 7 shows the analysis of Line 2 of Pipe 3.

B-19

Figure 9: Adept Pro Display of Defect Line 2. Adept Pro’s function is primarily to determine defect depth. Defect length and width are best obtained from axial and circumferential scans across the defect. Remote field eddy current signals spread in both the axial and the circumferential directions. To get length and width

Weld

Defects

B-20

requires corrections for the spread. Axial lengths estimated from the data should be reasonable. However, the combination of much greater spread in the circumferential direction combined with sensor separation means circumferential precision is poor. Pipe 2 was analyzed with an internally written MATLAB program. The fundamental equations are the same as used by Russell‘s Adept Pro software but there are some differences in the calibration. This can lead to small differences in the results for this pipe. This approach was used because Pipe 2 has two calibration defects with different depths. We expect the new calibration to give better results over a wide range of defect depths.

B-21

Table 1: C-scan Plots of defects found on Pipe 3 Test Line 1.

P3-10 P3-09

P3-07 P3-05

P3-04 P3-03

B-22

Table 2: Line 1 Defects Defect

Location

Max Depth (%

of wall thick.)

Defect

Length

Defect

Width

106” 96% 3” 2”

143.5” 78% 1.25” 1”

190” 92% 3.5” 2”

275.75” 75% 2” 2”

306.5” 84% 1.25” 1”

335” 87% 2” 1.5”

C-Scan plots The C-scan plots for all found defects are attached as a separate document. The tables containing Pipe 1 defects show the strip chart and C-scan for the phase only. The tables containing Pipe 2 defects show the C-scan for the phase only. Finally, the tables containing Pipe 3 defect information show the strip chart and C-scan for both the phase and magnitude. Summary Results Table The Excel spreadsheet summarizing the results is attached as a separate document. Pipe 2 data was only analyzed for the deepest pit. Data from Pipes 1 and 3 that showed dual pits are recorded in the spreadsheet as two measurements representing the maximum depth of each pit.

B-23


P1-12 P1-09

P1-07 P1-05

P1-04 P1-03

B-24

P1-02

B-25


P1-23 P1-22

P1-21 P1-18

P1-14 P1-13

B-26


P2-10 P2-09

P2-07 P2-06

P2-04

B-27


P2-20 P2-17

P2-14 P2-12

B-28


P3-10 P3-09

P3-07 P3-05

P3-04 P3-03

B-29


P3-23 P3-21

P3-19 P3-18

P3-17 P3-14

B-30

P3-12

B-31

BATTELLE DEMONSTRATION TEST DATA

B-32


B-33

Defect Number




Total Length of Metal Loss Region

Width of Metal Loss RegionMaximum Depth of Metal Loss Region


Comments


Defect Number




Total Length of Metal Loss Region

Width of Metal Loss RegionMaximum Depth of Metal Loss Region


Comments


184" to 196"

160" to 172"

188.8 192.2

TEST LINE 1

TEST DATA


PIPE SAMPLE 1

Benchmarking of Inspection TechnologiesDetection of Metal Loss - Page 1

Sensor Design:

Name:Date:Company:

Bruce NestlerothJanuary 26,2006Battelle

Rotating Permanent Magnet Eddy Current Inspection System

0.110

0.050

Various depths up to 0.150 inches

TEST LINE 2

1.0 0.075

Greater than 5 inches

0.165

1.5

1.0 0.075

120" to 144"

P1-1 328" to 340"

1.0

1.75

P1-4 256" to 268"

P1-23 74" to 86"

103.1 106.3

P1-3 280" to 292"


P1-2 304" to 316"

0.055

P1-5

0.115

Various depths up to 0.150 inches

No Metal Loss Detected

Yes. Raw Signals

12.0 126.0 138.0



P1-12 52" to 64"

P1-22 98" to 110"

P1-9 120" to 144"

P1-8

P1-6 208" to 220"

P1-7


Yes. Raw Signals

Yes. Raw Signals

P1-18 210" to 222" 2.0 0.155 Yes. Raw Signals212.5 216.8 3.3

0.75 0.020 Yes. Raw SignalsP1-17 234" to 246"

Yes. Raw SignalsP1-16 258" to 270"


P1-11 76" to 88" No Metal Loss Detected

1.75 58.1 62.3 3.2

WELD 180"

1.25 0.115


Small single pit

P1-10 100" to 112"

P1-20 160" to 172" No Metal Loss Detected

12.0 32.0

1.5 2.4

2.2


120.0 132.0


Yes. Raw Signals

Yes. Raw Signals

Yes. Raw Signals

Yes. Raw Signals

P1-13

Yes. Raw Signals

Yes. Raw Signals

Yes. Raw Signals

Yes. Raw Signals

WELD 120"

P1-14 306" to 318"

WELD 180"

232" to 244"

P1-15 282" to 294"

P1-19 186" to 198"

P1-21

WELD 120"


Yes. Raw Signals

Yes. Raw Signals

Yes. Raw Signals

Yes. Raw Signals


A area of general corrosion of variable depth that spans most sensors. A large wide corrosion area at 128".

A slow change in signal in all sensors throughout the region indicates a material property change

Yes. Raw Signals

Yes. Raw Signals





A large area of general corrosion of variable depth that spans the entire sensor width. Thecorrosion is close to the weld, altering both signals. A large wide corrosion area at 128"A slow change in signal in all sensors throughout the region indicates a

material property change

Yes. Raw Signals

Yes. Raw Signals

Yes. Raw Signals334.2 338.4 3.2 330" to 342" 1.0 0.075



233.2 237.0 2.8

260.1 264.9 3.8

288.0 292.8 3.8

74.8 79.1 3.3

241.1 241.8 0.7

306.9 310.8 2.9


B-34

B-35

B-36


Battelle – Rotating Magnetic Field InspectionJanuary 2006 Pipe 1 - Page 1

Pipe 1

Raw data output on same scale420 inches, 2 welds @ 120 and 180 inches

Search RegionExtra datafor noise

assessment

Extra datafor noise

assessment

348

354

360

366

372

CCW 2.5”CCW 2”CCW 1.5”CCW 1”CCW 0.5”0 degCW 0.5”CW 1”CW1.5”CW 2”CW 2.5


Axi

al S

enso

rs

Rad

ial S

enso

rs

Cal 1-1

Distance (inches)

Sen

sor O

utpu

t


322

328

334

340

346

P1-1

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

298

304

310

316

322

P1-2

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P1-3

274

280

286

292

298

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P1-4

250

256

262

268

274

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P1-5

226

232

238

244

250

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P1-6

202

208

214

220

226

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P1-7

178

184

190

196

202

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

Weld Signal

P1-8

Distance (inches)

Sen

sor O

utpu

t15

4

160

166

172

178



Axi

al S

enso

rs

Rad

ial S

enso

rs

Weld Signal


P1-9 note reporting area larger, 120 to 144 inches

114

120

126

132

138

144

150

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P1-10

94 100

106

112

118

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

Weld Signal


P1-11

70 76 82 88 94

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P1-12

46 52 58 64 70

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


324

330

336

342

348

P1-13

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

300

306

312

318

324

P1-14

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


276

282

288

294

300

P1-15

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

252

258

264

270

276

P1-16

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


228

234

240

246

252

P1-17

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

204

210

216

222

228

P1-18

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


180

186

192

198

204

P1-19

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

154

160

166

172

178

P1-20

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


114

120

126

132

138

144

150

P1-21 note reporting area larger, 120 to 144 inches

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

Weld Signal

92 98 104

110

116

P1-22

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

Weld Signal


68 74 80 86 92

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P1-23


Pipe 2

Raw data output on same scale360 inches,1 weld @120 inches


assessment

Extra datafor noise

assessment

Cal 2-1

292

298

304

310

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


Cal 2-2

264

270

276

282

288

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-1

288

294

300

306

312

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-2

264

270

276

282

288

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-3

240

246

252

258

264

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-4

216

222

228

234

240

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-5

192

198

204

210

216

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-6

168

174

180

186

192

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-7

144

150

156

162

168

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-8

120

126

132

138

144

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

Weld Signal

P2-Weld

108

114

120

126

132

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-9 Signal


P2-9

96 102

108

114

120

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-10

72 78 84 90 96

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-11

48 54 60 66 72

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-12

240

246

252

258

264

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-13

216

222

228

234

240

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-14

192

198

204

210

216

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-15

168

174

180

186

192

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-16

144

150

156

162

168

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-17

120

126

132

138

144

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

Weld Signal

P2-18

96 102

108

114

120

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


P2-19

72 78 84 90 96

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs

P2-20

48 54 60 66 72

Distance (inches)

Sen

sor O

utpu

t



Axi

al S

enso

rs

Rad

ial S

enso

rs


Pipe 3

Raw data output on same scale480 inches, 1 weld @ 240 inches


assessment

Extra datafor noise

assessment

Cal 3-1

409

415

421

427

433



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-1

378

384

390

396

402



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-2

354

360

366

372

378



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-3

324

330

336

342

348



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-4

294

300

306

312

318



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-5

264

270

276

282

288



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-Weld

228

234

240

246

252



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-6

216

222

228

234

240



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-7

180

186

192

198

204



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-8

156

162

168

174

180



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-9

132

138

144

150

156



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-10

96 102

108

114

120



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-11

60 66 72 78 84CCW 2.5”CCW 2”CCW 1.5”CCW 1”CCW 0.5”0 degCW 0.5”CW 1”CW1.5”CW 2”CW 2.5


Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-12

384

390

396

402

408



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-13

350

356

362

368

374



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-14

324

330

336

342

348



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-15

300

306

312

318

324



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-16

276

282

288

294

300



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-17

242

248

254

260

266



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-Weld

228

234

240

246

252



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-18

204

210

216

222

228



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-19

174

180

186

192

198



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-20

150

156

162

168

174



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-21

120

126

132

138

144



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

P3-22

96 102

108

114

120



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t


P3-23

60 66 72 78 84



Axi

al S

enso

rs

Rad

ial S

enso

rs

Distance (inches)

Sen

sor O

utpu

t

B-76


B-77

PACIFIC NORTHWEST NATIONAL LABORATORY (PNNL) DEMONSTRATION TEST DATA

B-78


B-79

B-80


B-81

B-82

0

10

20

30

40

50

60

70

0 50 100 150 200 250 300 350 400 450Distance (inches)

Ultr

ason

ic A

mpl

itude

-0.10

0.00

0.10

0.20

0.30

0.40

0.50

Ultr

ason

ic th

ickn

ess

inde

pend

ent m

easu

rem

ent

Amplitude

2 Cylindrical

1 Spherical

1Spherical

2 3 1Spherical

2 3 2 3 0

No dent

Ultrasonic birefringence

PNNL Ultrasonic measurements along the axis on Pipe 2, R Defects, at 15 degrees (approximately 3”) from TDC

B-83

OAKRIDGE NATIONAL LABORATORY (ORNL) DEMONSTRATION TEST DATA

B-84


B-85

Pipe Sample: 993Calibration

Crack Location

Length DepthMeasured

LengthMeasured

Depth

inches from end B inches

% wall thickness

186.4 2.558.7 586.4 582.4 2.544.4 3

Defect Number


End B)


Side B

End of Crack Region

from Side B






TEST LINE 1

NoneSCC2 226" to 242"

NoneSCC1 242" to 254"

NoneSCC3 210" to 222" 214" 216"

Another isolated at 142"SCC5 140" to 152" 145" 148"

None

Defect Set:893

Comments

1 multiple cracks; max = ~3/4"

multiple cracks; max = ~1/2"multiple cracks; max = ~1/2"

23

TEST DATA

Type of SCC

45

Blank Area:

Comments

26" Diameter Pipe with Stress Corrosion Cracks; Length = 27 feetPipe Sample:

CALIBRATION DATA


Sensor Design:

Name:

Date:Company:

Venugopal K. Varma, Austion Albrught, and Philip Bingham1/27/2006

Oak Ridge National Laboratory

EMAT shear Horizontal wave design

multiple cracks; max = ~1/4"multiple cracks; max = ~3 1/4"

SCC4 175" to 187" None

B-86

B-87

B-88


B-89

Calibration Note For calibration of SCC, a 26” pipe was provided with five SCC’s. These were located using liquid fluorescent magnetic particle inspection method. During the week of the testing we used the liquid fluorescent magnetic particle inspection to relocate the defects and had a hard time locating them. SCC 4 and SCC 5 could not be located and SCC3 and SCC 2 were indistinguishable from the scratches surrounding them. We could make out something SCC 2 &# area, but could not be confirmed. We cleaned the area using a wire brush and cleaner, but could not definitely identify the region having SCC. Only SCC1 could easily be identifiable, but this is more likely a manufacturing defect than an SCC. Due to lack of credible calibration data on 26 “ pipe, we had to base all algorithms on a previous 30” diameter training set. Venu, Philip, and Austin 1/27/2006

B-90


B-91

NATIONAL ENERGY TECHNOLOGY LABORATORY (NETL) DEMONSTRATION TEST DATA

B-92


B-93

Calibration Defect Location

Volume of Defect Depth of Defect Diameter of Defect Comments

inches from end A cubic inches inches inchesC1: 18 0.028 0.25 0.375

Defect Number

Search Region (Distance from End

A)

Location of Defect Region from Side A

Significance of Defect (Output/Calibration Output)

Volume of Defect (in3) (provided to


Depth of Defect (in) (provided to participant

after defect signif reported)Comments

inches inches cubic inches inches

None

52.55 1.31

D6 56" to 62" None

48" to 56"

D2

D4 42" to 48" 45.62

D3 34" to 42" None

Volume = 0.028

28" to 34" None

D1 18" to 28" 18.14 & 25.06 18.14" = 1, 25.06"=1.38

For significance: defect calibration hole @ 18" = 1 Vol @ 18" = 0.028 , Vol @ 25.06 = 0.039

CALIBRATION DATA

Defect

LINE 1

TEST DATA

6" Diameter, 0.5" Wall Thickness Pipe Sample, ~13' in lengthPipe Sample:Pipe Parameters:

PLASTIC PIPE SAMPLE

Benchmarking of Inspection TechnologiesDetection of Plastic Pipe Defects - Page 1

Sensor Design:

Name:Date:Company:

Jim Spenik, Chris Condon, Bill Fincham, Travis KirbySubmitted 01/23/06NETL

Capacitive sensor for Polyethylene Pipe Inspection

0.99

D5

D7 62" to 70" 66.36 1.15

Volume= 0.037

Volume = 0.033

D8 70" to 76" None

D9 76" to 84" None

D12 98" to 104" 101.03 1.61 Volume = 0.045

D11 90" to 98"

Volume = 0.02D13 104" to 112" 107.84 0.71

Volume = 0.016D14 112" to 118" 114.75 0.57

Volume = 0.020D15 118" to 126" 121.89 0.74We have indications that a consistant amount of material may have

been removed along the entire lengthD16 126" to 132" None ?

D17 132" to 138" None ?

D19 144" to 150"146.76 0.71

D10 84" to 90" 87.15 0.43

Volume = 0.020

We have indications that a consistant amount of material may have been removed along the entire length

D18 138" to 144" 138.3 1.13 Volume = 0.032

Volume = 0.012

B-94


APPENDIX C – DEVELOPER COMMENTS

C-1

SOUTHWEST RESEARCH INSTITUTE (SWRI) COMMENTS ON PIPELINE INSPECTION TECHNOLOGIES

DEMONSTRATION REPORT

C-2

This page intentionally blank

C-3

Final Comments on “PIPELINE INSPECTION TECHNOLOGIES

DEMONSTRATION REPORT”

APPLICATION OF REMOTE-FIELD EDDY CURRENT (RFEC) TESTING TO INSPECTION OF UNPIGGABLE PIPELINES

OTHER TRANSACTION AGREEMENT DTRS56-02-T-0001 SwRI® PROJECT 14.06162

PIPELINE AND HAZARDOUS MATERIALS SAFETY ADMNISTRATION U.S. DEPARTMENT OF TRANSPORTATION

SOUTHWEST RESEARCH INSTITUTE®

February 2006

Southwest Research Institute (SwRI) believes that the results of the demonstration testing indi-cate that the SwRI RFEC system is very promising as an inspection tool that can accurately detect and characterize wall-loss defects in pipelines. The report showed a comparison of pre-dicted vs. measured defect parameters with error bands of ±10% of wall thickness for defect depth and ±0.5 inch for defect length and depth. For the SwRI data, 68% of the predicted depths, 88% of the predicted lengths, and 88% of the predicted widths were within those error bands. If the error band is increased to ±20%, then 91% of the predicted depths would be within the band. The depth prediction had a systematic error in that the predicted depths were generally less than the measured ones. If corrections are made to the SwRI depth prediction algorithm to reduce the systematic error (for example, by using the demonstration test defect responses to correct the calibration approach), then even better results can be obtained.

It is emphasized that the SwRI RFEC tool was designed to meet the specifications and con-straints of the Explorer II robot under development by Carnegie Mellon University (as discussed in the SwRI comments on page B–4 of this report). The demonstration tests were thus conducted with sensors, instrumentation, data processing, scan speeds, etc. that are very representative of a field inspection system as integrated with Explorer II. SwRI therefore expects that results similar to those obtained in this demonstration would be obtained with an actual inspection system and that no degradation in performance would be experienced by transitioning to field hardware and inspection conditions.

C-4

Additional Information on the SwRI Remote Field Eddy Current Technology and Design as Integrated with the

Explorer II Robotic Platform SwRI Remote-Field Eddy Current – Through funding support from PHMSA/OPS, Southwest Research Institute® has developed a remote-field eddy current (RFEC) technology to be used in unpiggable lines. The SwRI RFEC tool is capable of detecting corrosion on the inside or outside pipe surface. Since a large percentage of pipelines cannot be inspected using “smart pig” tech-niques because of diameter restrictions, pipe bends, and valves, a concept for a collapsible excitation coil was developed but found unnecessary for the pipe sizes and materials of interest in this demonstration. A breadboard system that meets the size, power, and communication requirements for integration into the Carnegie Mellon Explorer II robot was developed and used in the demonstration tests. This system is shown in Figure 1. The demonstration system incor-porates eight detectors, and data from all eight channels are acquired and processed simultane-ously as the system is scanned along the pipe at speeds up to 4 inch/sec. All of the instrumenta-tion, except for a DC power supply and a laptop computer (used for storage of the processed data), is located on the tool. Figure 2 shows the system design as integrated with the Explorer II robot under development by Carnegie Mellon University. The RFEC system can expand to inspect 6- or 8-inch-diameter pipe and can retract to 4 inches to pass through obstructions.

Laptop Computer with CAN Bus Interface

Encoder Wheel

Electronics Sensors Excitation Coil

DC Power Supply

Figure 1. SwRI RFEC tool used in demonstration tests

C-5

Detector Module

Explorer Support Module

Exciter Module

Figure 2. SwRI RFEC tool design as integrated with Explorer II robot:

Top–Expanded for inspection with cover removed from exciter module, Bottom–Retracted to pass through restricted areas.

C-6


C-7

GAS TECHNOLOGY INSTITUTE (GTI) COMMENTS ON PIPELINE INSPECTION TECHNOLOGIES


C-8


C-9

Comments on the Comparison of Benchmarks and GTI Results

Albert Teitsma, Julie Maupin, and Paul Shuttleworth

Gas Technology Institute 1700 S. Mount Prospect Rd.

Des Plaines, IL 60018

25 October 2004 Introduction During the week of 9 January 2006, GTI staff came to the West Jefferson facility of Battelle Labs in Columbus, OH to test a prototype RFEC inspection vehicle in 3 sections of 8 inch pipe. We reported on our test results in a previous document.6 In this document we comment on the benchmarks reported in “Pipeline Inspection Technologies Demonstration Report” by Stephanie A. Flamberg and Robert C. Gertler. Comparison of Benchmarks and GTI Results Table 1 below compares GTI results to the benchmark data. There are two types of error in these results, systematic and random. The systematic errors are the average readings in Table 1, while scatter gives the random error. A different researcher analyzed the data from each pipe and the subjective components of the data analysis do show. All three underestimated the defect lengths, in one case by half an inch with a scatter of .4 inches. Particularly for small deep defects, this is too large an error, but the table also shows that proper analysis does give an acceptable precision (average=-0.139”, scatter=0.133). Precision in the circumferential direction was not as good, but as pointed out in a previous report, remaining strength calculation such as B31G or RSTRENG do not use circumferential extent in the calculations.

Figure 1. Data with Pipe 3 corrected for calibration error.

There was a serious depth calibration error for pipe three, which made the scatter for the GTI results look worse than it was. Figure 1 shows the improvement with recalibrated data. GTI expected that the anticipated error would be about +/- 10% of the full wallthickness, as indicated by the lines in Figure1. Table 1 shows that more experienced analysts can achieve that, the scatter for Pipe 1 being 10%, while that for Pipe 3 was a mere 7%. GTI’s sizing of the natural corrosion areas was excellent.

6 “Analysis of Sensor Benchmarking Tests: Remote Field Eddy Current Technique”, Julie Maupin, Albert Teitsma, and Paul Shuttleworth.

C-10

Time to Take the Data Since time to take the data has become an issue, GTI has included results from its run with Russell NDE Systems, Inc. equipment, which GTI plans to use in its modules, in this report.

Figure 2. A faster run using Russell NDE Systems, Inc. instrumentation.

GTI inspected 23’ of Pipe 3 in 7 minutes using this instrumentation, which GTI brought along for demonstration purposes only. The speed was limited by the speed of our tow motor. The instrumentation can easily handle the 4” per second specified for Explorer II. The unfiltered data in Figure 2 is a little noisier than that obtained from the laboratory lock-in amplifier, but more than good enough for the size of the signals obtained during the benchmark tests. GTI concentrated on maximizing signal strength and minimizing power consumption. Speed at the very low speeds used by Explorer II was never an issue. For most of the measurements, it took GTI a little over half a day per run in Pipes 1 and 5, and a little longer in Pipe 3 using a single lock-in amplifier to measure all sixteen channels. To ensure superior data quality the lock-in was allowed to settle nearly a second before reading the data from a sensing coil. C-Scans C-scans obtained with the RFEC inspection do not have the resolution of the benchmark scans, but the correlation between them are excellent. Figure 3 compares the natural corrosion defect, P1-23. Similar results are obtained for the other defects.

C-11

Figure 3. Excellent correlation between the RFEC results and the natural corrosion benchmark data. Conclusion The results clearly demonstrate that the RFEC technique is eminently suited for inspecting transmission and distribution piping. The measurements had excellent quality. However GTI’s analysis indicates that it takes experienced analysts to translate the measurements into precise defect severity estimates. Although most of the results were not obtained at inspection speeds, the short run with more realistic field equipment showed that inspection at Explorer II speeds will not reduce the quality of the defect severity measurements.

C-12

C-13

BATTELLE COMMENTS ON PIPELINE INSPECTION TECHNOLOGIES


C-14


C-15

Comments on Demonstration Results for the

ROTATING PERMANENT MAGNET

INSPECTION TOOL Prepared by Battelle

February 17, 2006

Theory of Operation

The rotating permanent magnet inspection method employed by Battelle at the Pipeline Inspection Technologies Demonstration is an alternative to the common concentric coil methods to induce low-frequency eddy currents in ferromagnetic pipe and tubes. Battelle’s technology consists of a pair of permanent magnets that rotate around a central axis in proximity to the inner surface of the pipe sample. The rotating permanent magnet pairs are used to induce high current densities in the material undergoing inspection. Following fundamental laws of electrical induction, rotating permanent magnet pairs inside a pipe along its longitudinal axis establishes an alternating electrical current in the wall of the pipe. Figure 1, a cutaway drawing showing the rotating permanent magnet exciter, illustrates this concept. The current flows in an elliptical path around the magnets. When the magnetizer is vertical, strong currents flow axially along the top and bottom of the pipe and circumferentially at the sides. When the magnetizer is horizontal, strong currents flow circumferentially at the sides of the pipe and axially at the top and the bottom. Finite element modeling shows that a two-pole magnetizer produces strong current densities at distances reasonably far away from the magnetizer. Although the current is complex at the magnet poles (where it is strongest), at distances of a pipe diameter or more away from the magnetizer it is uniform and sinusoidal. With this uniform energy induced in the pipe, simple magnetic field sensors can be used to detect the change in current densities in the pipe wall and thus pinpoint the location of defects and anomalies. The development of this technology began in fall 2003 and is sponsored by The U.S. Department of Energy’s National Energy Technology Laboratory with cofunding from the Pipeline Research Council International. The first known use of this inspection method to detect corrosion was performed in September 2004.

C-16

Figure 2. Illustration of the rotating permanent magnet exciter and sensor location

System Configuration as Demonstrated

Figure 2 shows the prototype used for the 8 inch corrosion inspection benchmark demonstration. A pair of NdFeB magnets is mounted on a steel core machined from 1018 steel. The magnets are 2 inches long, 1 inch wide, and 0.5 inch thick; the magnet strength is 38 MegaGauss-Oersted. While the strong holding force secures the magnets on the steel core, copper covers keep the magnets precisely aligned. The air gap between the magnet and the pipe wall is 0.5 inch. Wheeled support plates keep the magnet centered in the pipe. A variable speed direct current motor is used to rotate the magnetizing assembly. The rotational speed used in this demonstration was 300 rpm or 5 Hz. The power required to rotate the magnets at this speed was about 70 watts. While this is above the available power of 50 watts budgeted by Explorer II, this power requirement is significantly better than the 200 watts required in prior designs. Three pairs of axial and a radial Hall Effect sensors were mounted in 4 sensor shoes designed to ride on the ID of the pipe. While sensor to magnet spacing of 8 to 10 inches provides stronger signal changes from corrosion anomalies, the distance from the magnet to the sensor was 13 inches to meet EXPLORER II specifications. To continuously monitor rotational speed, a small magnet was attached to the shaft and an additional Hall Effect sensor was used to produce a synchronous signal.

C-17

Figure 3. Rotating permanent Magnet Inspection system as configured for the technology demonstration A 24 channel real-time data recorder system was implemented and fundamental experiments were conducted to provide data to aid in the design of the rotating magnetizer. A system was designed to simultaneously record and process 11 sensor pairs, the sync signal and one open channel. The block diagram of the data recorder system is shown in Figure 3. The heart of the recorder is the National Instruments PXI-4472, an eight-channel dynamic signal acquisition module for making high-accuracy frequency-domain measurements. The eight NI PXI-4472 input channels simultaneously digitize input signals over a bandwidth from 0.5 Hz to 45 kHz. Three PXI-4472 modules were synchronized to provide 24 channel input using the PXI chassis and a star trigger bus. The PXI chassis communicates with a desktop computer using a fiber optic link. The desktop computer is used to analyze the signals using a lock-in amplifier approach, as described in a previous DOE semiannual report. LabVIEW software modules for lock-in amplifier measurements were used in the development of a custom data acquisition and display program.

Figure 3. The block diagram of the data acquisition system

Display of results

C-18

The data acquisition and processing hardware and software processed signals and displayed data in real time during the demonstration. A typical output of the data recording package is shown in Figure 4. In real time display mode, the data scrolls along the monitor as the inspection tool traverses the pipe. The upper and lower graphs show the axial and radial sensors respectively using a stacked line plotting routine, a format familiar to pigging vendors and users of pipeline inspection technologies. In this figure, the signal from an axially short, circumferentially wide metal loss anomaly can be seen in the middle channels of each sensor type.

Figure 4. Screen capture of custom LabVIEW data acquisition and display program The results submitted by Battelle on January 26, 2006 (contained in appendix B) included signals from each reporting area in a uniform format. An example signal is shown in Figure 5 for pipe sample 2, search area 10. The upper and lower stacked graphs show the signals from the axial and radial sensors respectively; the color codes repeat so that sensor pairs can be correlated. Since only about 70 degrees of the pipe was instrumented, the center sensor was positioned so that it traversed the centerline of the defect. In some of the graphs in appendix B it is evident that the tool rotated slightly as it was pulled through the pipe because some of the corrosion signals are greater in other sensors. The signals provided with the report were plotted on the same scale for quick visual comparison. For detection and assessment, signals were amplified so that smaller corrosion areas could be more easily detected and assessed. Other graphical representations, including plotting axial versus radial signals, are proving to be useful in assessing corrosion. A scaled topographical map of the corrosion depth is included at the bottom of Figure 5 after it was flipped (the tool was pulled from right to left). The two humps in the stacked graphs correspond to the two pits in the image. In the reported results, the presence of single or multiple pits was indicated in the comment section. The depth assessment was based on the largest signal since the data reporting form specifically requested maximum depth.

C-19

Figure 5. Signal from pipe sample 2, search area 10

Comments on Results

The results presented in the main section of the demonstration report are representative of the current capability of the rotating permanent magnet tool. This comparatively new inspection methodology is in its third year of development. Specific comments on detection and sizing results are provided next. Detection. The results of the demonstration showed that all corrosion anomalies were detected and one additional anomaly was falsely detected. The false call anomaly was assessed as small and not detected in all pulls. The spacing between sensors (sensor pitch) of the demonstration configuration was 0.5 inches. For corrosion with shallow depth and a width and length nominally the same as the sensor pitch, a detectable signal may only be produced by a sensor traveling directly underneath the anomaly. Two sensors straddling the same anomaly may not produce a signal. Future implementations may need a finer sensor pitch to improve results. Corrosion sizing. A corrosion anomaly locally increases the density of the currents that are induced by the rotating magnetizer. The local change in current density is also influenced by the length and width. The algorithm for estimating the depth of the corrosion anomaly includes these three measures, in a manner similar to magnetic flux leakage data analysis methods. Data from the calibration anomalies and the first benchmark demonstration were used to establish the sizing algorithm. The unity plot shown in the main report indicates a good correlation between measured and predicted values, however there is a general tendency to under-call the depth. This was the first algorithm developed for corrosion anomaly depth assessment. Additional data and algorithm refinement should help improve results.

C-20

Natural Corrosion Sample. The natural corrosion sample was difficult to assess because of the unexpected weld. In hindsight, the signals were quite clear. Figures 6 and 7 show the reported raw data with new annotations for lines 1 and 2 respectively. In the results reported on January 27, 2006 for these lines Battelle discussed:

• Line 1 - A large area of general corrosion of variable depth that spans the entire sensor width. The corrosion is close to the weld, altering both signals. A large wide corrosion area at 128"

• Line 2 - An area of general corrosion of variable depth that spans most sensors. A large wide corrosion area at 128"

The signal 128 inches from the end was the unexpected weld signal. The general corrosion on either side of the weld corresponds to the measured results; however the close welds caused interference and sizing was not attempted at this time. While the natural corrosion pipe was complex, it is only one of many unique challenges that must be faced when implementing inspection technology and the experience will be valuable in future developments.

Summary

The benchmarking results are a representative assessment of the current state of development of the rotating permanent magnet inspection system. The planned improvements of this technology should advance the capability of this inspection system. Battelle is currently working on reducing magnetizer size, increasing rotation speed, and increasing the separation distance between the magnet and the pipe. Separations of over an inch appear to be practical, which will aid in the implementation of this technology. The rapid advances of this new inspection technology should make this methodology useful for unpigable pipeline applications in the near future.

Acknowledgement

The U.S. Department of Energy’s National Energy Technology Laboratory is sponsoring this development with cofunding from the Pipeline Research Council International. However, any opinions, findings, conclusions, or recommendations expressed herein are those of the authors and do not necessarily reflect the views of these sponsors.

C-21

Figure 6. Natural corrosion results line 1

Figure 7. Natural corrosion results, line 2

C-22


C-23

PACIFIC NORTHWEST NATIONAL LABORATORY (PNNL) COMMENTS ON PIPELINE INSPECTION TECHNOLOGIES


C-24


C-25

Comments on NETL and Pipeline & Hazardous Materials Safety Administration pipeline inspection technologies demonstration

Submitted by: Paul D. Panetta

Pacific Northwest National Laboratory Richland, WA 99352

[email protected] (509) 372-6107

The Pacific Northwest National Laboratory (PNNL) participated in the Pipeline Inspection Technologies Demonstration during the week of January 9, 2006. The main focus of the demonstration was to rank the severity of dents based on ultrasonic measurements of the mechanical properties and the presence of plastic strain. This approach is dramatically different than the current assessment based solely on dimensional measurements. The advantage of this approach is that the reliability of the pipeline can be determined based on material properties and how they change with time and damage, rather than the size and shape of a dent. Measurements were performed on two 24 inch diameter pipes containing dents and dents with gouges. Pipe 1 contained 3 rows of dents from a track how with a very small separation distance, on the order of a few inches in some cases. The total number of dents exceeded 40 dents. The operation of creating these dents and dents with gouges created a significant amount of distortion to the pipe and ovalization of the pipe. In-service pipelines with the amount of denting are highly unlikely and this pipe does not represent a realistic pipeline operating scenario. Despite this significant distortion results were promising. Pipe 2 contained 10 dents and 11 reporting locations. All dents were successfully detected and estimates of the size were provided. The ultrasonic strain measurement correctly ranked 7 out of the 9 reporting locations for 100% detectability and 77% accuracy on ranking severity. The sensor was a non contact electromagnetic acoustic transducer (EMAT) that was scanned along the axis of the pipe at several distances from the dents placed at top dead center. The sensor and cart are shown in Figure 1. Figure 2 shows the amplitude and ultrasonic measurements along pipe 2 with the sensor placed 15 degrees from top dead center. The amplitude clears shows a deviation at each reporting location with a dent and no deviation where there is no dent. The ultrasonic shear wave birefringence is independent of thickness which is critical for characterizing mechanical properties due to deformation because a simple thickness measurement is NOT an accurate assessment of strain. The inspection speed was as fast as 5 inches per second and the electronics can operate as fast as 4 or 5 feet per second (~3 MPH). The measurements were performed in a 24 inch pipe and are amenable to pipes as small as 4 inches in diameter. The technology proved to be very sensitive to mechanical damage due to dents and is also ideal for application where pipelines are bent due to subsidence or other earth movement. This technology is ready for incorporation onto robotics platforms and for field testing and subsequent commercialization for specific applications.

C-26

EMAT Sensor

Motor for sensor rotation

Springs for smooth motion past dents

Figure 1. Photo of the ultrasonic sensor and scanning cart.

0

10

20

30

40

50

60

70

0 50 100 150 200 250 300 350 400 450Distance (inches)

Ultr

ason

ic A

mpl

itude

-0.10

0.00

0.10

0.20

0.30

0.40

0.50

Ultr

ason

ic th

ickn

ess

inde

pend

ent m

easu

rem

ent

Amplitude

2 Cylindrical

1 Spherical

1Spherical

2 3 1Spherical

2 3 2 3 0

No dent

Ultrasonic birefringence

PNNL Ultrasonic measurements along the axis on Pipe 2, R Defects, at 15 degrees (approximately 3”) from TDC

Figure 2. Amplitude and ultrasonic birefringence as a function of distance along pipe 2.

C-27

OAKRIDGE NATIONAL LABORATORY (ORNL) COMMENTS ON PIPELINE INSPECTION TECHNOLOGIES


C-28


C-29

SCC detection using Shear Horizontal EMAT

Based on the results, we feel that the ORNL SCC detection system using shear horizontal wave EMAT detection has performed very well for the test conditions. In this response, we address areas pertaining to: training data issues, lack of data on SCC depth, additional defects along test line 1, and false positives. With these comments concentrating on issues where the results are in question, we would like to emphasis that the system performed very well for the test and addressing these issues will only improve or better clarify the results. Training Data: The current ORNL set-up for detecting SCCs with the shear horizontal wave EMAT uses transmitted signals to assess the presence of a crack. The signals from ‘no-flaw’ regions are compared to the signals from ‘flaw’ regions to identify cracks. The key issue in performing this measurement is the determination of features, derived from the response signals that separate flawed regions from those with no flaws. In the current algorithm, wavelet based features from both ‘flaw’ and ‘no-flaw’ regions are used to establish classes (SCC, no-flaw, other anomaly). Since this technology requires training data of known defect and no-defect regions, a 26” training pipe was provided in addition to the test pipe during our visit to the test facility. Unfortunately, we were unable to generate a proper training set from this test pipe due to the quality and the discrepancy in the location of the flaws. Instead, previous data collected from a 30” pipe for training were used. Although the mode frequencies were different for the 26" and 30" pipe due to change in wall thickness and pipe diameter the results were still satisfactory. This indicates a robustness of the training sets across pipe diameters and thicknesses. The system performance would have only improved had we used a training set generated out of similar pipe geometry. SCC Depth Data: Defect sizes were given in terms of length and area of crack on the pipe with no depth information. Liquid fluorescent magnetic particle inspection for detecting SCCs does not contain any information on the depth of the crack, while the EMAT based approach has a direct dependence on it. Hence, some very small cracks detected by magnetic particle method may not be detected by EMAT due to their depth being small. This is a possible reason for SSC2 not being detected. With the knowledge of SCC depth, we could have determined how well the system is able to detect the severity of the crack. Additional Known Defects in Test Line 1: In testing, we were instructed to test along three different lines of the test pipe to determine the presence of defects over particular spans along each line. Figure 1 shows the pipe layout for the test. Each test box (blue boxes labeled SCC1 – SCC14) along with every defect previously identified on the pipe (pink boxes labeled 4-9, 15-20) are pictured. The dashed lines represent the three scan lines. As mentioned in the results, the SCC defect we were to locate in SCC3 is defect 8 (far left side of box). However, we positioned this defect to the right by several inches. Since the EMATs scans an arc of ~12 inches around the circumference of the pipe, the SCC boxes within the figure have been drawn with 12 inch height to show the area covered by the sensor. From the figure, we see that defects 17, 18, and 19 are all on the upper edge of SCC3.

C-30

Figure 4. Test setup.

Our defect detection signal is essentially a distance measurement from the no-defect class within our feature space. This distance is pictured in Figure 2 for the SCC3 region. Red lines show boundary of box SCC3 and the approximate locations of defects 8, 17, 18, and 19 are shown in pink text. In our response to the test, we listed the defect in the SCC3 box based on the large signal that appears to correspond to defect 18 (a fairly large inclusion). From the signal, we do feel that we are seeing the intended defect 8 as well but did not list it due to its location straddling the boundary of the SCC3 region.

C-31

Figure 5. Defect response signal for SCC3 area.

False Positives: As mentioned in the results, we also identified a false positive on each scan line. The EMATs did indicate flaws in areas where they were none, and this could be the result of not having the baseline data or the algorithm needing further refinement. Lack of good natural SCC data has been one of the difficulties we faced while developing this technology. We have created synthetic SCCs using electrical discharge machining (EDM), however, EDM machined SCCs do not give a signature truly characteristic of a natural SCC. Figure 3 shows the signals returned for the three false positives that have peaks similar to our previous experience with SCC signatures. Red lines delineate the regions of interest. The false positive on line 1 (Figure 3a) shows a series of peaks each similar in shape to an SCC response. The false positive in line 2 (Figure 3b) shows a well-isolated peak typical of an SCC response. The false positive in line 3 (Figure 3c) shows an SCC type response on the right side of SCC14 box. Similar bumps also can be seen near 160" mark but were not marked as SCCs due to the low dome shape of the response.

C-32

(a) (b)

(c)

Figure 6. False positive signals for scan lines 1, 2 and 3 are shown in (a), (b) and (c) respectively

Conclusion: As mentioned earlier, we feel the ORNL SCC detection system performed well in this test. Lessons learned from the tests are: 1) Training data may not be necessary for each pipe geometry being investigated, and 2) Information on SCC depth is needed to fully characterize the system performance. We feel that the system performance will continue to improve as more training data from natural SCCs are collected and used to train the algorithm.

C-33

NATIONAL ENERGY TECHNOLOGY LABORATORY (NETL) COMMENTS ON PIPELINE INSPECTION TECHNOLOGIES


C-34


C-35

Analysis of Sensor Benchmark Tests Capacitive Sensor for Polyethylene Pipe Inspection

Prepared by: James Spenik, Chris Condon, Bill Fincham, Travis Kirby

National Energy Technology Laboratory 3610 Collins Ferry Road Morgantown, WV 26505

February, 15, 2006

C-36


Introduction: Representatives from the National Energy Technology Laboratory demonstrated polyethylene pipe inspection technology at Battelle’s West Jefferson Pipeline Simulation Facility near Columbus, OH. The technology was demonstrated January 10 – 12, 2006 by James Spenik (REM), Chris Condon (REM), Bill Fincham (Parsons) and Travis Kirby (WVU). Battelle provided a 13-foot length of 6-inch nominal diameter, 0.5-inch wall thickness polyethylene pipe. Holes and saw cuts were placed into the top outer surface of the pipe along an axial line. Twelve defects were placed within nineteen 6-inch long search regions. Eight of the regions did not contain a defect, one region contained two defects. The line of defects was covered thus the experimenters did not know their location when data was collected. However, a calibration defect was available whose characteristics and location was known to the experimenters. The probe was able to identify the defect in every search region without false positives. Technique: Abnormalities in the pipe wall are determined by changes in the dielectric properties of the wall material. An electric field is projected through the pipe wall by the probe head (Fig. 1). The wall material behaves as the dielectric component of a capacitor. This arrangement formed the probe head of the sensor device. Since the dielectric constant of polyethylene is greater than that of air (or natural gas) an absence of material within the electric field will manifest itself as a decrease in capacitance.

Figure 1 Projection of Electric Field through Pipe Wall The probe head and associated electronics were mounted on a platform designed for this particular test (Fig. 2). The probe head was mounted 5.5 inches from the back circular disk of the 9.25-inch long platform. A 5.5 inch diameter disk was mounted at each end of the platform. In future use, the probe could be incorporated on existing platforms. The platform was propelled through the pipe using a stationary stepper motor and nylon filament. An optical encoder was used to determine probe position within the pipe. Data were transmitted using RF transmission via Bluetooth technology. Another option would be to store the data onboard and retrieved at a later time. Power was supplied using an on-board 9-volt

Dielectric Material (wall)

Probe head

Field LinesDefect

battery. The data transmission rate for this particular demonstration was controlled by the optical encoder and stepper motor. Capacitance data were to be transmitted every forty counts of the optical encoder (0.09 inch axial movement) but this value may have varied a few counts. The stepper motor moved the platform at a rate of approximately 0.09 in/s. The sampling rate was approximately 1 Hz for this configuration due to the constraints previously mentioned. Thus the transit time through the pipe was approximately 15 minutes. However, the electronics package used is capable of transmission rates of between 45 – 90 Hz and modifications to the package would allow transmission rates in the MHz range.

Figure 2 Platform with Probe Electronics

Data collection/analysis: Twenty traverses were performed during the three days of data acquisition. The first ten were preliminary to identify problems. These difficulties were not related to the function of the pipe defect sensor but rather sensor movement. Initially, the optical encoder did not react to movement along the surface of the yellow polyethylene pipe. This was an unforeseen problem since, in an earlier test, the encoder reacted in black polyethylene pipe. The problem was resolved by placing a strip of material visible to the encoder on the interior lower surface of the pipe. Movement of the platform would be halted due to a slightly underpowered stepper motor. The edges of the platform disks were lubricated with graphite which minimized the problem. The deviation of the probe head from a linear path was minimized using guide line attached to the bottom of the pipe and through the bottom of both platform disks. These problems were identified and solved during the first ten traverses. Data from the second set of ten traverses were useful and provided data for statistical analysis. Tests commenced with the rear disk approximately 1.5 inch from the “B” end of the pipe placing the sensor head at the 149 inch position of the 156 inch (13 foot) pipe (Fig. 3). Tests concluded with the probe head at the 7 inch position. Run11 – Run 20 were compiled to determine the position of anomalies within the polyethylene pipe.

Probe Head

Figure 3 Path of Probe Through Pipe The average number of data points accumulated for each run was 1619 points corresponding to a measurement for each 0.0877 inches of travel. However, this number varied between runs with a standard deviation of 46. These discrepancies can be attributed to either binding of the stepper motor or variations in the triggering level of the optical sensor. The focus of the research was creation of the probe; the platform was designed only after conformation of the teams’ participation in the demonstration was received. Data were post processed to determine the exact position of each defect within each search region in the following manner:

1. Data were aligned so the minimum capacitance for each run near the calibration hole coincided. (Minimum capacitance corresponds to the center of the anomaly)

2. Since the total length of the traverse and the total number of data points were known, the

ratio of these numbers yielded an initial estimate of step size for each run. 3. The data for each run was separated by search region.

4. The position and value of the minimum capacitance value within the search region for

each run was determined.

5. The average position of the minimum within the region was determined.

6. Each run was realigned within the region so the minimum was located at the average minimum position.

This method was effective; however, cumulative error caused the position of the anomaly within a search region to be progressively misinterpreted. The measured position and actual position of the defect in search area D1 was at 25 inch, however, the actual position of the defect in search area D19 was 148” and the measured position was 146.8” Again, this is not due to sensor error but rather due to positioning error. All defects were identified with the exception of a binary defect (two holes separated by 0.5 inch on centers) located at position D15 which we identified as a single entity. The probe in its current configuration was not designed to separate binary anomalies separated by less than an inch. Although it was not part of the benchmarking demonstration, an attempt was made to provide a comparative value of volume of material removed by the defect. Only moderate success was achieved in this endeavor. The reason that definitive volumetric values could not be determined was because the defects presented in the pipe could be considered to have three variables: diameter, depth and type (round hole or saw cut). Due to the nature of the electric field produced

A B Traverse Direction

by the probe, the depth and diameter cannot be combined into the single variable volume. Since the electric field strength diminishes as a function of distance from the probe head, a smaller volume closer to the probe head is seen as equivalent to a larger volume further away. The output from the current probe design yields two values: capacitance and change in capacitance with respect to axial position. Therefore there were three unknowns and only two equations and thus the volume of material removed was indeterminate. A future design of the probe allowing circumferential measurements will allow the development of an algorithm to define defect volume. Figures 4 through 7 illustrate the typical probe response when a defect was encountered. Each figure compiles the ten runs taken within an eight-inch long region of interest. The abscissa is the variation from minimum capacitance within the region and the ordinate is the linear position. Figure 4 shows the calibration defect and the probe response. Figure 5 shows a typical response to a round defect and Figure 6 indicates the response to a saw cut. Figure 7 indicates the probe response in a region with no known defects. The presence of an anomaly typically produced a variation of 4000 aF. Variations in a region without defined anomalies were typically 500 aF. Conclusion: The probe successfully identified the position of all defects within the search regions and had no false positive results. Deviations from the precise position of the defects within the search region can be attributed to the means of locomotion and position identification procedures. The data acquisition rate can be markedly increased with a superior locomotion scheme. Further devlopment to this technology will produce a device that can be inserted into in situ natural gas pipelines and determine their integrity.

25 - 17 (18 - 24) min @ 18.14

0

10000

20000

30000

40000

50000

60000

70000

17 18 19 20 21 22 23 24 25

Run 11Run 12Run 13Run 14Run 15Run 16Run 17Run 18Run 19bRun 20

Figure 4 Calibration Hole and Probe Response (18” position)

31 - 23 (24 - 30) min @ 25.06

0

10000

20000

30000

40000

50000

60000

70000

23 24 25 26 27 28 29 30 31


Figure 5 Round Hole Defect and Probe Response (25” position)

103 - 95 (96 - 102) min @ 101.03

0

10000

20000

30000

40000

50000

60000

70000

95 96 97 98 99 100 101 102 103


Figure 6 Saw Cut Defect and Probe Response (102” position)

Figure 7 Probe Response with no Defect (102” position)

85 - 77 (78 - 84)

0

10000

20000

30000

40000

50000

60000

70000

77 78 79 80 81 82 83 84 85


Date post:	15-Aug-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Pipeline & Hazardous Materials Safety Administration · 2018-03-23 · The PHMSA Pipeline Safety...

Documents